commit fa38774a1e050bf2c1ce0f1b7561e61884eb9b55 Author: ModelHub XC Date: Sun Sep 20 14:58:13 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: OpenDataArena/Qwen2.5-7B-ODA-Mixture-500k Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..e475b8c --- /dev/null +++ b/.gitattributes @@ -0,0 +1,64 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text + + +*.bz2 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 +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack 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 +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +saved_model/**/* 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 +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zstandard filter=lfs diff=lfs merge=lfs -text +*.tfevents* filter=lfs diff=lfs merge=lfs -text +*.db* filter=lfs diff=lfs merge=lfs -text +*.ark* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text + +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.gguf* filter=lfs diff=lfs merge=lfs -text +*.ggml filter=lfs diff=lfs merge=lfs -text +*.llamafile* filter=lfs diff=lfs merge=lfs -text +*.pt2 filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text + +checkpoint-6870/training_args.bin filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/vocab.json filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/merges.txt filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/tokenizer.json filter=lfs diff=lfs merge=lfs -text +model-00001-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +merges.txt filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/model-00004-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +training_args.bin filter=lfs diff=lfs merge=lfs -text +model-00004-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +vocab.json filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/model-00001-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/model-00002-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +checkpoint-6870/model-00003-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +model-00003-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text +model-00002-of-00004.safetensors filter=lfs diff=lfs merge=lfs -text \ No newline at end of file diff --git a/README.md b/README.md new file mode 100644 index 0000000..eab3267 --- /dev/null +++ b/README.md @@ -0,0 +1,249 @@ +--- +base_model: Qwen/Qwen2.5-7B-Base +library_name: transformers +pipeline_tag: text-generation +datasets: +- OpenDataArena/ODA-Mixture-500k +tags: +- qwen2.5 +- sft +- opendataarena +- oda-mixture-500k +license: apache-2.0 +language: +- en +metrics: +- accuracy +--- + +# Qwen2.5-7B-ODA-Mixture-500k +Leaderboard Performance + +Qwen2.5-7B-ODA-Mixture-500k is a supervised fine-tuned (SFT) model built on top of **Qwen2.5-7B-Base**, trained with **[ODA-Mixture-500k](https://huggingface.co/datasets/OpenDataArena/ODA-Mixture-500k)**. This training set is curated by mixing top-performing open corpora selected via the *[OpenDataArena](https://opendataarena.github.io)* leaderboard, and refined through deduplication and benchmark decontamination, aiming to improve the model’s general capabilities across **General**, **Math**, **Code**, and **Reasoning** domains. + +--- + +## 🧠 Model Summary + +- **Base Model**: `Qwen/Qwen2.5-7B-Base` +- **Training Data**: `OpenDataArena/ODA-Mixture-500k` +- **Domain Coverage**: General, Math, Code, Reasoning +- **Scale (selected training set)**: ~**500K** samples +- **Goal**: Improve multi-domain reasoning and problem-solving ability while achieving stable overall performance gains. + +--- + +## ⚙️ Training Data Curation Pipeline + +ODA-Mixture-500k is built by following a single rule: **trust the OpenDataArena leaderboard**. + +### 1️⃣ Data Collection + +We utilize **LIMO** as the foundational anchor due to its exceptional sample efficiency on the ODA overall leaderboard. To scale up to 500K, we integrate the top-performing and efficient corpora from specific ODA domain leaderboards, including: +- **AM-Thinking-v1-Distilled-Math** (Math domain) +- **AM-Thinking-v1-Distilled-code** (Code domain) +- **math-gpt-4o-200k** (General domain) +- **SYNTHETIC-2-SFT-verified** (Reasoning domain) + +### 2️⃣ Deduplication & Decontamination + +We first perform **exact deduplication** over all questions to remove identical items, and then run **benchmark decontamination** to reduce evaluation leakage by removing overlaps with standard and competition benchmarks. + +### 3️⃣ Data Selection + +At the 500K scale, our priority is distributional coverage. We employ semantic clustering to partition the total data pool into distinct thematic clusters. Within each cluster, we perform uniform sampling to ensure the final mixture represents a broad and balanced spectrum of reasoning tasks, maximizing the model's generalization capabilities. + +--- + +## 📚 Training Data Source Composition + +| Source | Count | Percentage | +|---|---:|---:| +| LIMO | 817 | 0.16% | +| AM-Thinking-Distilled-math | 150,244 | 29.67% | +| AM-Thinking-Distilled-code| 150,252 | 29.67% | +| math-gpt-4o-200k| 100,138 | 19.78% | +| SYNTHETIC-2-SFT-verified| 104,913 | 20.72% | + +--- + +## 🧩 Data Format + +The training data sample format is as follows (aligned with the dataset schema): + +```json +{ + "id": "unique_identifier", + "source": "data source", + "question": "textual question or instruction", + "response": "textual response" +} +``` + +--- + +## 📈 Performance + +Qwen2.5-7B-ODA-Mixture-500k is evaluated as an SFT model built on **Qwen2.5-7B-Base** across the full ODA benchmark suite spanning four domains: +- **General (DROP, IFEVAL, AGIEVAL, MMLU-Pro)** +- **Math (GSM8K, MATH500, Omni-Math, OlympiadBench, AIME2024)** +- **Code (HumanEval, MBPP, LCB (V5), HumanEval+)** +- **Reasoning (ARC-C, BBH, CALM, KOR-BENCH)**. + +We observe consistent improvements over the base checkpoint, with particularly strong gains on several benchmarks. + +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+ Leaderboard Performance Comparison. Best scores in bold, second-best underlined. Eff. denotes Data Efficiency. +
Model / Training DataSizeEff.GeneralMathCodeReasoningAVG
Qwen2.5-7B-Base
Qwen2.5-7B-Base--51.439.850.142.746.0
OpenThoughts3-1.2M1.2M+0.01145.571.867.054.359.6
OmniThought-0528365k+0.02747.171.247.657.255.8
SYNTHETIC-2-SFT-verified105k+0.08651.369.840.158.955.0
AM-Thinking-v1-Distilled-math558k+0.01657.777.439.544.854.8
LIMO817+9.92060.744.057.953.854.1
MiroMind-M1-SFT-719K719k+0.00652.071.026.351.550.2
AM-Thinking-v1-Distilled-code324k+0.02449.952.368.744.453.8
Light-R1-SFTData79k+0.08455.564.438.851.952.7
ODA-Mixture-500k500k+0.03963.472.866.759.665.6
ODA-Mixture-101k101k+0.14956.871.264.451.561.0
+
+ +--- + +## 🌐 About OpenDataArena + +[OpenDataArena](https://opendataarena.github.io/) is an open research platform dedicated to **discovering, evaluating, and advancing high-quality datasets for AI post-training**. It provides a transparent, data-centric ecosystem to support reproducible dataset evaluation and sharing. + +**Key Features:** +- 🏆 **Dataset Leaderboard** — helps researchers identify **the most valuable and high-quality datasets across different domains** +- 📊 **Detailed Evaluation Scores** — provides **comprehensive metrics** to assess data quality, complexity, difficulty, etc. +- 🧰 **Data Processing Toolkit** — [OpenDataArena-Tool](https://github.com/OpenDataArena/OpenDataArena-Tool) offers an open-source pipeline for dataset curation and scoring. + +--- + +## 🚀 Usage + +Model repo: `OpenDataArena/Qwen2.5-7B-ODA-Mixture-500k`. Below is a minimal runnable example for loading and inference: + +```python +from transformers import AutoModelForCausalLM, AutoTokenizer + +MODEL_ID = "OpenDataArena/Qwen2.5-7B-ODA-Mixture-500k" + +tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True) +model = AutoModelForCausalLM.from_pretrained(MODEL_ID, device_map="auto", trust_remote_code=True) + +messages = [ + {"role": "user", "content": "Natalia sold clips to 48 of her friends in April, and then she sold half as many clips in May. How many clips did Natalia sell altogether in April and May?"}, +] +text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) +inputs = tokenizer([text], return_tensors="pt").to(model.device) + +outputs = model.generate( + **inputs, + max_new_tokens=256, + do_sample=True, + temperature=0.7, + top_p=0.9, +) +print(tokenizer.decode(outputs[0], skip_special_tokens=True)) +``` + +--- + +## 📚 Citation + +If you use this model or its training data (ODA-Mixture-500k), please cite: + +```bibtex +@article{gao2025closing, + title={Closing the Data Loop: Using OpenDataArena to Engineer Superior Training Datasets}, + author={Gao, Xin and Wang, Xiaoyang and Zhu, Yun and Cai, Mengzhang and He, Conghui and Wu, Lijun}, + journal={arXiv preprint arXiv:2601.09733}, + year={2025} +} +``` +```bibtex +@article{cai2025opendataarena, + title={OpenDataArena: A Fair and Open Arena for Benchmarking Post-Training Dataset Value}, + author={Cai, Mengzhang and Gao, Xin and Li, Yu and Lin, Honglin and Liu, Zheng and Pan, Zhuoshi and Pei, Qizhi and Shang, Xiaoran and Sun, Mengyuan and Tang, Zinan and others}, + journal={arXiv preprint arXiv:2512.14051}, + year={2025} +} +``` \ No newline at end of file diff --git a/README.md.backup b/README.md.backup new file mode 100644 index 0000000..0cdff66 --- /dev/null +++ b/README.md.backup @@ -0,0 +1,60 @@ +--- +library_name: transformers +license: other +base_model: /mnt/dhwfile/raise/user/caimengzhang/huggingface/hub/Qwen2.5-7B +tags: +- llama-factory +- full +- generated_from_trainer +model-index: +- name: seed-42 + results: [] +--- + + + +# seed-42 + +This model is a fine-tuned version of [/mnt/dhwfile/raise/user/caimengzhang/huggingface/hub/Qwen2.5-7B](https://huggingface.co//mnt/dhwfile/raise/user/caimengzhang/huggingface/hub/Qwen2.5-7B) on the overall-v4c-500k dataset. + +## Model description + +More information needed + +## Intended uses & limitations + +More information needed + +## Training and evaluation data + +More information needed + +## Training procedure + +### Training hyperparameters + +The following hyperparameters were used during training: +- learning_rate: 5e-05 +- train_batch_size: 1 +- eval_batch_size: 8 +- seed: 42 +- distributed_type: multi-GPU +- num_devices: 32 +- total_train_batch_size: 32 +- total_eval_batch_size: 256 +- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments +- lr_scheduler_type: cosine +- lr_scheduler_warmup_ratio: 0.1 +- num_epochs: 3.0 + +### Training results + + + +### Framework versions + +- Transformers 4.55.0 +- Pytorch 2.6.0+cu124 +- Datasets 3.2.0 +- Tokenizers 0.21.0 diff --git a/added_tokens.json b/added_tokens.json new file mode 100644 index 0000000..482ced4 --- /dev/null +++ b/added_tokens.json @@ -0,0 +1,24 @@ +{ + "": 151658, + "": 151657, + "<|box_end|>": 151649, + "<|box_start|>": 151648, + "<|endoftext|>": 151643, + "<|file_sep|>": 151664, + "<|fim_middle|>": 151660, + "<|fim_pad|>": 151662, + "<|fim_prefix|>": 151659, + "<|fim_suffix|>": 151661, + "<|im_end|>": 151645, + "<|im_start|>": 151644, + "<|image_pad|>": 151655, + "<|object_ref_end|>": 151647, + "<|object_ref_start|>": 151646, + "<|quad_end|>": 151651, + "<|quad_start|>": 151650, + "<|repo_name|>": 151663, + "<|video_pad|>": 151656, + "<|vision_end|>": 151653, + "<|vision_pad|>": 151654, + "<|vision_start|>": 151652 +} diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000..5a41134 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 3.0, + "total_flos": 1.440651216420864e+16, + "train_loss": 0.30719185047274594, + "train_runtime": 200812.4774, + "train_samples_per_second": 1.094, + "train_steps_per_second": 0.034 +} \ No newline at end of file diff --git a/chat_template.jinja b/chat_template.jinja new file mode 100644 index 0000000..d153f82 --- /dev/null +++ b/chat_template.jinja @@ -0,0 +1,4 @@ +{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ 'System: ' + system_message + '<|endoftext|>' + ' +' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '<|endoftext|>' + ' +Assistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + ' +' }}{% endif %}{% endfor %} \ No newline at end of file diff --git a/checkpoint-6870/added_tokens.json b/checkpoint-6870/added_tokens.json new file mode 100644 index 0000000..482ced4 --- /dev/null +++ b/checkpoint-6870/added_tokens.json @@ -0,0 +1,24 @@ +{ + "": 151658, + "": 151657, + "<|box_end|>": 151649, + "<|box_start|>": 151648, + "<|endoftext|>": 151643, + "<|file_sep|>": 151664, + "<|fim_middle|>": 151660, + "<|fim_pad|>": 151662, + "<|fim_prefix|>": 151659, + "<|fim_suffix|>": 151661, + "<|im_end|>": 151645, + "<|im_start|>": 151644, + "<|image_pad|>": 151655, + "<|object_ref_end|>": 151647, + "<|object_ref_start|>": 151646, + "<|quad_end|>": 151651, + "<|quad_start|>": 151650, + "<|repo_name|>": 151663, + "<|video_pad|>": 151656, + "<|vision_end|>": 151653, + "<|vision_pad|>": 151654, + "<|vision_start|>": 151652 +} diff --git a/checkpoint-6870/chat_template.jinja b/checkpoint-6870/chat_template.jinja new file mode 100644 index 0000000..d153f82 --- /dev/null +++ b/checkpoint-6870/chat_template.jinja @@ -0,0 +1,4 @@ +{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ 'System: ' + system_message + '<|endoftext|>' + ' +' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '<|endoftext|>' + ' +Assistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + ' +' }}{% endif %}{% endfor %} \ No newline at end of file diff --git a/checkpoint-6870/config.json b/checkpoint-6870/config.json new file mode 100644 index 0000000..fa2559a --- /dev/null +++ b/checkpoint-6870/config.json @@ -0,0 +1,59 @@ +{ + "architectures": [ + "Qwen2ForCausalLM" + ], + "attention_dropout": 0.0, + "bos_token_id": 151643, + "eos_token_id": 151643, + "hidden_act": "silu", + "hidden_size": 3584, + "initializer_range": 0.02, + "intermediate_size": 18944, + "layer_types": [ + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention" + ], + "max_position_embeddings": 131072, + "max_window_layers": 28, + "model_type": "qwen2", + "num_attention_heads": 28, + "num_hidden_layers": 28, + "num_key_value_heads": 4, + "rms_norm_eps": 1e-06, + "rope_scaling": null, + "rope_theta": 1000000.0, + "sliding_window": null, + "tie_word_embeddings": false, + "torch_dtype": "bfloat16", + "transformers_version": "4.55.0", + "use_cache": false, + "use_mrope": false, + "use_sliding_window": false, + "vocab_size": 152064 +} diff --git a/checkpoint-6870/generation_config.json b/checkpoint-6870/generation_config.json new file mode 100644 index 0000000..7bc5028 --- /dev/null +++ b/checkpoint-6870/generation_config.json @@ -0,0 +1,6 @@ +{ + "bos_token_id": 151643, + "eos_token_id": 151643, + "max_new_tokens": 2048, + "transformers_version": "4.55.0" +} diff --git a/checkpoint-6870/merges.txt b/checkpoint-6870/merges.txt new file mode 100644 index 0000000..80c1a19 --- /dev/null +++ b/checkpoint-6870/merges.txt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5 +size 1671853 diff --git a/checkpoint-6870/model-00001-of-00004.safetensors b/checkpoint-6870/model-00001-of-00004.safetensors new file mode 100644 index 0000000..85f508e --- /dev/null +++ b/checkpoint-6870/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b7eed591ff517e9f22b7e6d51bc9b93a4aea6d18841ec2070224d80d19c523a +size 4877660776 diff --git a/checkpoint-6870/model-00002-of-00004.safetensors b/checkpoint-6870/model-00002-of-00004.safetensors new file mode 100644 index 0000000..ecb8bc2 --- /dev/null +++ b/checkpoint-6870/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2d23ea5f1df6344f99df01328398513600a92c66f133cb3d3a759d267348e75 +size 4932751008 diff --git a/checkpoint-6870/model-00003-of-00004.safetensors b/checkpoint-6870/model-00003-of-00004.safetensors new file mode 100644 index 0000000..5fed81c --- /dev/null +++ b/checkpoint-6870/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ceafc6ab27e5fdceea8b39ed6f307f85d847b6090a88ee188de5c09e1fd852bc +size 4330865200 diff --git a/checkpoint-6870/model-00004-of-00004.safetensors b/checkpoint-6870/model-00004-of-00004.safetensors new file mode 100644 index 0000000..6bf211e --- /dev/null +++ b/checkpoint-6870/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c944547784945651653accbcd4142dbc94e5ddcc47d6ea9bcf07b220b96720c9 +size 1089994880 diff --git a/checkpoint-6870/model.safetensors.index.json b/checkpoint-6870/model.safetensors.index.json new file mode 100644 index 0000000..098a4a8 --- /dev/null +++ b/checkpoint-6870/model.safetensors.index.json @@ -0,0 +1,347 @@ +{ + "metadata": { + "total_parameters": 333312, + "total_size": 15231233024 + }, + "weight_map": { + "lm_head.weight": "model-00004-of-00004.safetensors", + "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00003-of-00004.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00003-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-00003-of-00004.safetensors", + "model.layers.18.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00003-of-00004.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.k_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.q_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.v_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00003-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.q_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.v_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00003-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00003-of-00004.safetensors", + "model.layers.27.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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00001-of-00004.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-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-00001-of-00004.safetensors", + "model.layers.7.self_attn.k_proj.bias": "model-00001-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.bias": "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.bias": "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.self_attn.k_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.norm.weight": "model-00003-of-00004.safetensors" + } +} diff --git a/checkpoint-6870/special_tokens_map.json b/checkpoint-6870/special_tokens_map.json new file mode 100644 index 0000000..17305b3 --- /dev/null +++ b/checkpoint-6870/special_tokens_map.json @@ -0,0 +1,31 @@ +{ + "additional_special_tokens": [ + "<|im_start|>", + "<|im_end|>", + "<|object_ref_start|>", + "<|object_ref_end|>", + "<|box_start|>", + "<|box_end|>", + "<|quad_start|>", + "<|quad_end|>", + "<|vision_start|>", + "<|vision_end|>", + "<|vision_pad|>", + "<|image_pad|>", + "<|video_pad|>" + ], + "eos_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-6870/tokenizer.json b/checkpoint-6870/tokenizer.json new file mode 100644 index 0000000..51ebb3b --- /dev/null +++ b/checkpoint-6870/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa +size 11421896 diff --git a/checkpoint-6870/tokenizer_config.json b/checkpoint-6870/tokenizer_config.json new file mode 100644 index 0000000..e812d6b --- /dev/null +++ b/checkpoint-6870/tokenizer_config.json @@ -0,0 +1,208 @@ +{ + "add_bos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "151643": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151644": { + "content": "<|im_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151645": { + "content": "<|im_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151646": { + "content": "<|object_ref_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151647": { + "content": "<|object_ref_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151648": { + "content": "<|box_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151649": { + "content": "<|box_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151650": { + "content": "<|quad_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151651": { + "content": "<|quad_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151652": { + "content": "<|vision_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151653": { + "content": "<|vision_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151654": { + "content": "<|vision_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151655": { + "content": "<|image_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151656": { + "content": "<|video_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151657": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151658": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151659": { + "content": "<|fim_prefix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151660": { + "content": "<|fim_middle|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151661": { + "content": "<|fim_suffix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151662": { + "content": "<|fim_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151663": { + "content": "<|repo_name|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151664": { + "content": "<|file_sep|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "additional_special_tokens": [ + "<|im_start|>", + "<|im_end|>", + "<|object_ref_start|>", + "<|object_ref_end|>", + "<|box_start|>", + "<|box_end|>", + "<|quad_start|>", + "<|quad_end|>", + "<|vision_start|>", + "<|vision_end|>", + "<|vision_pad|>", + "<|image_pad|>", + "<|video_pad|>" + ], + "bos_token": null, + "clean_up_tokenization_spaces": false, + "eos_token": "<|endoftext|>", + "errors": "replace", + "extra_special_tokens": {}, + "model_max_length": 131072, + "pad_token": "<|endoftext|>", + "padding_side": "right", + "split_special_tokens": false, + "tokenizer_class": "Qwen2Tokenizer", + "unk_token": null +} diff --git a/checkpoint-6870/trainer_state.json b/checkpoint-6870/trainer_state.json new file mode 100644 index 0000000..1f1f2d9 --- /dev/null +++ b/checkpoint-6870/trainer_state.json @@ -0,0 +1,9652 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 6870, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002183406113537118, + "grad_norm": 15.530065061351076, + "learning_rate": 2.9112081513828244e-07, + "loss": 3.5144, + "step": 5 + }, + { + "epoch": 0.004366812227074236, + "grad_norm": 14.048089091944986, + "learning_rate": 6.550218340611353e-07, + "loss": 3.3991, + "step": 10 + }, + { + "epoch": 0.006550218340611353, + "grad_norm": 10.38275753075565, + "learning_rate": 1.0189228529839884e-06, + "loss": 3.4492, + "step": 15 + }, + { + "epoch": 0.008733624454148471, + "grad_norm": 9.374594765056237, + "learning_rate": 1.3828238719068415e-06, + "loss": 3.3321, + "step": 20 + }, + { + "epoch": 0.010917030567685589, + "grad_norm": 6.58125566549559, + "learning_rate": 1.7467248908296944e-06, + "loss": 3.0597, + "step": 25 + }, + { + "epoch": 0.013100436681222707, + "grad_norm": 7.80412598343367, + "learning_rate": 2.1106259097525473e-06, + "loss": 2.9508, + "step": 30 + }, + { + "epoch": 0.015283842794759825, + "grad_norm": 7.524557299709741, + "learning_rate": 2.4745269286754004e-06, + "loss": 2.612, + "step": 35 + }, + { + "epoch": 0.017467248908296942, + "grad_norm": 6.2330781402437365, + "learning_rate": 2.8384279475982535e-06, + "loss": 2.3435, + "step": 40 + }, + { + "epoch": 0.019650655021834062, + "grad_norm": 5.3971059460100825, + "learning_rate": 3.2023289665211066e-06, + "loss": 1.9812, + "step": 45 + }, + { + "epoch": 0.021834061135371178, + "grad_norm": 5.173910417939837, + "learning_rate": 3.5662299854439593e-06, + "loss": 1.6557, + "step": 50 + }, + { + "epoch": 0.024017467248908297, + "grad_norm": 4.343854825431366, + "learning_rate": 3.930131004366813e-06, + "loss": 1.4334, + "step": 55 + }, + { + "epoch": 0.026200873362445413, + "grad_norm": 3.318343002698737, + "learning_rate": 4.2940320232896655e-06, + "loss": 1.2603, + "step": 60 + }, + { + "epoch": 0.028384279475982533, + "grad_norm": 3.3649234822213425, + "learning_rate": 4.657933042212519e-06, + "loss": 1.128, + "step": 65 + }, + { + "epoch": 0.03056768558951965, + "grad_norm": 4.651497105373944, + "learning_rate": 5.021834061135372e-06, + "loss": 1.0685, + "step": 70 + }, + { + "epoch": 0.03275109170305677, + "grad_norm": 2.1122195307286127, + "learning_rate": 5.385735080058224e-06, + "loss": 0.9865, + "step": 75 + }, + { + "epoch": 0.034934497816593885, + "grad_norm": 3.264835592266701, + "learning_rate": 5.749636098981077e-06, + "loss": 0.8563, + "step": 80 + }, + { + "epoch": 0.03711790393013101, + "grad_norm": 4.854625880384626, + "learning_rate": 6.11353711790393e-06, + "loss": 0.813, + "step": 85 + }, + { + "epoch": 0.039301310043668124, + "grad_norm": 2.2913931555435223, + "learning_rate": 6.477438136826783e-06, + "loss": 0.8189, + "step": 90 + }, + { + "epoch": 0.04148471615720524, + "grad_norm": 1.4161513553023397, + "learning_rate": 6.841339155749636e-06, + "loss": 0.7341, + "step": 95 + }, + { + "epoch": 0.043668122270742356, + "grad_norm": 1.2517988600799463, + "learning_rate": 7.205240174672489e-06, + "loss": 0.7264, + "step": 100 + }, + { + "epoch": 0.04585152838427948, + "grad_norm": 4.403128247316299, + "learning_rate": 7.569141193595343e-06, + "loss": 0.7105, + "step": 105 + }, + { + "epoch": 0.048034934497816595, + "grad_norm": 2.6065584518468996, + "learning_rate": 7.933042212518195e-06, + "loss": 0.6883, + "step": 110 + }, + { + "epoch": 0.05021834061135371, + "grad_norm": 1.2548695388960658, + "learning_rate": 8.296943231441049e-06, + "loss": 0.6444, + "step": 115 + }, + { + "epoch": 0.05240174672489083, + "grad_norm": 1.2420608105158033, + "learning_rate": 8.660844250363902e-06, + "loss": 0.6228, + "step": 120 + }, + { + "epoch": 0.05458515283842795, + "grad_norm": 1.6213174255897973, + "learning_rate": 9.024745269286754e-06, + "loss": 0.6091, + "step": 125 + }, + { + "epoch": 0.056768558951965066, + "grad_norm": 1.1139361754005903, + "learning_rate": 9.388646288209607e-06, + "loss": 0.6049, + "step": 130 + }, + { + "epoch": 0.05895196506550218, + "grad_norm": 2.019221687744237, + "learning_rate": 9.75254730713246e-06, + "loss": 0.6334, + "step": 135 + }, + { + "epoch": 0.0611353711790393, + "grad_norm": 1.4614530692241856, + "learning_rate": 1.0116448326055314e-05, + "loss": 0.5484, + "step": 140 + }, + { + "epoch": 0.06331877729257641, + "grad_norm": 1.7546701255165822, + "learning_rate": 1.0480349344978165e-05, + "loss": 0.5777, + "step": 145 + }, + { + "epoch": 0.06550218340611354, + "grad_norm": 1.5028564001758502, + "learning_rate": 1.084425036390102e-05, + "loss": 0.5427, + "step": 150 + }, + { + "epoch": 0.06768558951965066, + "grad_norm": 0.8737189391813837, + "learning_rate": 1.1208151382823872e-05, + "loss": 0.5501, + "step": 155 + }, + { + "epoch": 0.06986899563318777, + "grad_norm": 0.7913219748536747, + "learning_rate": 1.1572052401746727e-05, + "loss": 0.5404, + "step": 160 + }, + { + "epoch": 0.07205240174672489, + "grad_norm": 1.2931917942966242, + "learning_rate": 1.1935953420669578e-05, + "loss": 0.5317, + "step": 165 + }, + { + "epoch": 0.07423580786026202, + "grad_norm": 1.7815089914348687, + "learning_rate": 1.2299854439592432e-05, + "loss": 0.5391, + "step": 170 + }, + { + "epoch": 0.07641921397379912, + "grad_norm": 1.588354964609756, + "learning_rate": 1.2663755458515283e-05, + "loss": 0.5163, + "step": 175 + }, + { + "epoch": 0.07860262008733625, + "grad_norm": 0.7324777658190085, + "learning_rate": 1.3027656477438139e-05, + "loss": 0.5206, + "step": 180 + }, + { + "epoch": 0.08078602620087336, + "grad_norm": 0.9320526641008475, + "learning_rate": 1.339155749636099e-05, + "loss": 0.51, + "step": 185 + }, + { + "epoch": 0.08296943231441048, + "grad_norm": 1.0473996131197105, + "learning_rate": 1.3755458515283843e-05, + "loss": 0.4995, + "step": 190 + }, + { + "epoch": 0.0851528384279476, + "grad_norm": 1.1005598187154526, + "learning_rate": 1.4119359534206697e-05, + "loss": 0.4992, + "step": 195 + }, + { + "epoch": 0.08733624454148471, + "grad_norm": 1.3835272450195761, + "learning_rate": 1.448326055312955e-05, + "loss": 0.5157, + "step": 200 + }, + { + "epoch": 0.08951965065502183, + "grad_norm": 0.8363321954594061, + "learning_rate": 1.4847161572052404e-05, + "loss": 0.4884, + "step": 205 + }, + { + "epoch": 0.09170305676855896, + "grad_norm": 0.6751320611923934, + "learning_rate": 1.5211062590975255e-05, + "loss": 0.4837, + "step": 210 + }, + { + "epoch": 0.09388646288209607, + "grad_norm": 0.6597124878200359, + "learning_rate": 1.557496360989811e-05, + "loss": 0.4614, + "step": 215 + }, + { + "epoch": 0.09606986899563319, + "grad_norm": 1.0002616519579557, + "learning_rate": 1.5938864628820962e-05, + "loss": 0.5178, + "step": 220 + }, + { + "epoch": 0.0982532751091703, + "grad_norm": 2.094018394615991, + "learning_rate": 1.6302765647743815e-05, + "loss": 0.4785, + "step": 225 + }, + { + "epoch": 0.10043668122270742, + "grad_norm": 0.8762721338958381, + "learning_rate": 1.6666666666666667e-05, + "loss": 0.4867, + "step": 230 + }, + { + "epoch": 0.10262008733624454, + "grad_norm": 1.3388349700796345, + "learning_rate": 1.703056768558952e-05, + "loss": 0.4781, + "step": 235 + }, + { + "epoch": 0.10480349344978165, + "grad_norm": 1.0507336135636423, + "learning_rate": 1.7394468704512373e-05, + "loss": 0.4732, + "step": 240 + }, + { + "epoch": 0.10698689956331878, + "grad_norm": 0.7938940927456775, + "learning_rate": 1.7758369723435225e-05, + "loss": 0.4651, + "step": 245 + }, + { + "epoch": 0.1091703056768559, + "grad_norm": 1.0437122295269432, + "learning_rate": 1.812227074235808e-05, + "loss": 0.4585, + "step": 250 + }, + { + "epoch": 0.11135371179039301, + "grad_norm": 0.67350165285708, + "learning_rate": 1.848617176128093e-05, + "loss": 0.4609, + "step": 255 + }, + { + "epoch": 0.11353711790393013, + "grad_norm": 0.5808251314496068, + "learning_rate": 1.8850072780203784e-05, + "loss": 0.4549, + "step": 260 + }, + { + "epoch": 0.11572052401746726, + "grad_norm": 1.8396131242666172, + "learning_rate": 1.921397379912664e-05, + "loss": 0.4783, + "step": 265 + }, + { + "epoch": 0.11790393013100436, + "grad_norm": 0.9680443797016299, + "learning_rate": 1.9577874818049492e-05, + "loss": 0.455, + "step": 270 + }, + { + "epoch": 0.12008733624454149, + "grad_norm": 0.7014587496843943, + "learning_rate": 1.9941775836972345e-05, + "loss": 0.4479, + "step": 275 + }, + { + "epoch": 0.1222707423580786, + "grad_norm": 0.9218263081064745, + "learning_rate": 2.0305676855895198e-05, + "loss": 0.456, + "step": 280 + }, + { + "epoch": 0.12445414847161572, + "grad_norm": 0.5840696636129238, + "learning_rate": 2.066957787481805e-05, + "loss": 0.4318, + "step": 285 + }, + { + "epoch": 0.12663755458515283, + "grad_norm": 0.6901911501394837, + "learning_rate": 2.1033478893740903e-05, + "loss": 0.4494, + "step": 290 + }, + { + "epoch": 0.12882096069868995, + "grad_norm": 0.5256416173639158, + "learning_rate": 2.1397379912663756e-05, + "loss": 0.4604, + "step": 295 + }, + { + "epoch": 0.13100436681222707, + "grad_norm": 0.8643348227959773, + "learning_rate": 2.176128093158661e-05, + "loss": 0.4529, + "step": 300 + }, + { + "epoch": 0.1331877729257642, + "grad_norm": 0.845191908182343, + "learning_rate": 2.2125181950509464e-05, + "loss": 0.4462, + "step": 305 + }, + { + "epoch": 0.13537117903930132, + "grad_norm": 0.6911667544775351, + "learning_rate": 2.2489082969432314e-05, + "loss": 0.4353, + "step": 310 + }, + { + "epoch": 0.13755458515283842, + "grad_norm": 0.6842730240845932, + "learning_rate": 2.2852983988355166e-05, + "loss": 0.4425, + "step": 315 + }, + { + "epoch": 0.13973799126637554, + "grad_norm": 0.6095243486936803, + "learning_rate": 2.3216885007278022e-05, + "loss": 0.4375, + "step": 320 + }, + { + "epoch": 0.14192139737991266, + "grad_norm": 0.9176516550209551, + "learning_rate": 2.3580786026200875e-05, + "loss": 0.4539, + "step": 325 + }, + { + "epoch": 0.14410480349344978, + "grad_norm": 0.8430335294570548, + "learning_rate": 2.3944687045123728e-05, + "loss": 0.4347, + "step": 330 + }, + { + "epoch": 0.1462882096069869, + "grad_norm": 0.7088826431924614, + "learning_rate": 2.430858806404658e-05, + "loss": 0.4149, + "step": 335 + }, + { + "epoch": 0.14847161572052403, + "grad_norm": 0.6180099894880557, + "learning_rate": 2.4672489082969433e-05, + "loss": 0.422, + "step": 340 + }, + { + "epoch": 0.15065502183406113, + "grad_norm": 0.6817744275567881, + "learning_rate": 2.503639010189229e-05, + "loss": 0.4418, + "step": 345 + }, + { + "epoch": 0.15283842794759825, + "grad_norm": 0.6940745586404697, + "learning_rate": 2.540029112081514e-05, + "loss": 0.4404, + "step": 350 + }, + { + "epoch": 0.15502183406113537, + "grad_norm": 0.5754528111989647, + "learning_rate": 2.576419213973799e-05, + "loss": 0.4263, + "step": 355 + }, + { + "epoch": 0.1572052401746725, + "grad_norm": 1.0591998260395188, + "learning_rate": 2.6128093158660844e-05, + "loss": 0.4299, + "step": 360 + }, + { + "epoch": 0.15938864628820962, + "grad_norm": 0.7123628186533127, + "learning_rate": 2.64919941775837e-05, + "loss": 0.4196, + "step": 365 + }, + { + "epoch": 0.1615720524017467, + "grad_norm": 0.6457647967870623, + "learning_rate": 2.6855895196506553e-05, + "loss": 0.4174, + "step": 370 + }, + { + "epoch": 0.16375545851528384, + "grad_norm": 0.9935438036526899, + "learning_rate": 2.7219796215429405e-05, + "loss": 0.3993, + "step": 375 + }, + { + "epoch": 0.16593886462882096, + "grad_norm": 0.883212399644436, + "learning_rate": 2.7583697234352258e-05, + "loss": 0.4345, + "step": 380 + }, + { + "epoch": 0.16812227074235808, + "grad_norm": 0.948205577828456, + "learning_rate": 2.794759825327511e-05, + "loss": 0.4441, + "step": 385 + }, + { + "epoch": 0.1703056768558952, + "grad_norm": 0.8223542412825284, + "learning_rate": 2.831149927219796e-05, + "loss": 0.4029, + "step": 390 + }, + { + "epoch": 0.17248908296943233, + "grad_norm": 0.7182654852607071, + "learning_rate": 2.867540029112082e-05, + "loss": 0.4342, + "step": 395 + }, + { + "epoch": 0.17467248908296942, + "grad_norm": 0.6862331015738732, + "learning_rate": 2.9039301310043672e-05, + "loss": 0.4044, + "step": 400 + }, + { + "epoch": 0.17685589519650655, + "grad_norm": 0.9719904263631477, + "learning_rate": 2.940320232896652e-05, + "loss": 0.4114, + "step": 405 + }, + { + "epoch": 0.17903930131004367, + "grad_norm": 0.8428735007204966, + "learning_rate": 2.9767103347889374e-05, + "loss": 0.4124, + "step": 410 + }, + { + "epoch": 0.1812227074235808, + "grad_norm": 0.8900932432816406, + "learning_rate": 3.0131004366812227e-05, + "loss": 0.4096, + "step": 415 + }, + { + "epoch": 0.18340611353711792, + "grad_norm": 0.658732643364676, + "learning_rate": 3.0494905385735083e-05, + "loss": 0.4019, + "step": 420 + }, + { + "epoch": 0.185589519650655, + "grad_norm": 0.5717380212732919, + "learning_rate": 3.0858806404657935e-05, + "loss": 0.4125, + "step": 425 + }, + { + "epoch": 0.18777292576419213, + "grad_norm": 0.7561360040926198, + "learning_rate": 3.122270742358079e-05, + "loss": 0.3979, + "step": 430 + }, + { + "epoch": 0.18995633187772926, + "grad_norm": 0.7154647354779691, + "learning_rate": 3.158660844250364e-05, + "loss": 0.411, + "step": 435 + }, + { + "epoch": 0.19213973799126638, + "grad_norm": 1.106580393251376, + "learning_rate": 3.1950509461426493e-05, + "loss": 0.4001, + "step": 440 + }, + { + "epoch": 0.1943231441048035, + "grad_norm": 0.8277158276869705, + "learning_rate": 3.2314410480349346e-05, + "loss": 0.3982, + "step": 445 + }, + { + "epoch": 0.1965065502183406, + "grad_norm": 0.6695298962156193, + "learning_rate": 3.26783114992722e-05, + "loss": 0.4129, + "step": 450 + }, + { + "epoch": 0.19868995633187772, + "grad_norm": 0.8926693328948101, + "learning_rate": 3.304221251819505e-05, + "loss": 0.4173, + "step": 455 + }, + { + "epoch": 0.20087336244541484, + "grad_norm": 0.9134165249644083, + "learning_rate": 3.3406113537117904e-05, + "loss": 0.3947, + "step": 460 + }, + { + "epoch": 0.20305676855895197, + "grad_norm": 0.6780077341796262, + "learning_rate": 3.377001455604076e-05, + "loss": 0.4066, + "step": 465 + }, + { + "epoch": 0.2052401746724891, + "grad_norm": 0.8359819073368681, + "learning_rate": 3.413391557496361e-05, + "loss": 0.4054, + "step": 470 + }, + { + "epoch": 0.2074235807860262, + "grad_norm": 0.49801067958580947, + "learning_rate": 3.449781659388647e-05, + "loss": 0.3923, + "step": 475 + }, + { + "epoch": 0.2096069868995633, + "grad_norm": 0.7389933814419388, + "learning_rate": 3.486171761280932e-05, + "loss": 0.4068, + "step": 480 + }, + { + "epoch": 0.21179039301310043, + "grad_norm": 0.4893296535673842, + "learning_rate": 3.5225618631732174e-05, + "loss": 0.4002, + "step": 485 + }, + { + "epoch": 0.21397379912663755, + "grad_norm": 0.5093678162190949, + "learning_rate": 3.558951965065502e-05, + "loss": 0.4068, + "step": 490 + }, + { + "epoch": 0.21615720524017468, + "grad_norm": 0.9016225376861969, + "learning_rate": 3.595342066957787e-05, + "loss": 0.4081, + "step": 495 + }, + { + "epoch": 0.2183406113537118, + "grad_norm": 0.5844526927479975, + "learning_rate": 3.6317321688500726e-05, + "loss": 0.4053, + "step": 500 + }, + { + "epoch": 0.2205240174672489, + "grad_norm": 0.6328384104413857, + "learning_rate": 3.6681222707423585e-05, + "loss": 0.391, + "step": 505 + }, + { + "epoch": 0.22270742358078602, + "grad_norm": 0.6295475420727077, + "learning_rate": 3.704512372634644e-05, + "loss": 0.4061, + "step": 510 + }, + { + "epoch": 0.22489082969432314, + "grad_norm": 0.6065260581614148, + "learning_rate": 3.740902474526929e-05, + "loss": 0.4107, + "step": 515 + }, + { + "epoch": 0.22707423580786026, + "grad_norm": 1.004727626897767, + "learning_rate": 3.777292576419214e-05, + "loss": 0.3998, + "step": 520 + }, + { + "epoch": 0.2292576419213974, + "grad_norm": 1.1129157537413497, + "learning_rate": 3.8136826783114996e-05, + "loss": 0.4005, + "step": 525 + }, + { + "epoch": 0.2314410480349345, + "grad_norm": 0.9960094773975019, + "learning_rate": 3.850072780203784e-05, + "loss": 0.4098, + "step": 530 + }, + { + "epoch": 0.2336244541484716, + "grad_norm": 0.5262503731188081, + "learning_rate": 3.88646288209607e-05, + "loss": 0.3965, + "step": 535 + }, + { + "epoch": 0.23580786026200873, + "grad_norm": 0.6294913749088836, + "learning_rate": 3.9228529839883554e-05, + "loss": 0.3962, + "step": 540 + }, + { + "epoch": 0.23799126637554585, + "grad_norm": 0.6672651864348986, + "learning_rate": 3.9592430858806406e-05, + "loss": 0.3976, + "step": 545 + }, + { + "epoch": 0.24017467248908297, + "grad_norm": 0.6736374930092932, + "learning_rate": 3.995633187772926e-05, + "loss": 0.3932, + "step": 550 + }, + { + "epoch": 0.2423580786026201, + "grad_norm": 0.9025467677457151, + "learning_rate": 4.032023289665211e-05, + "loss": 0.4163, + "step": 555 + }, + { + "epoch": 0.2445414847161572, + "grad_norm": 0.6615595441983655, + "learning_rate": 4.068413391557497e-05, + "loss": 0.3885, + "step": 560 + }, + { + "epoch": 0.24672489082969432, + "grad_norm": 0.747795749818688, + "learning_rate": 4.104803493449782e-05, + "loss": 0.3979, + "step": 565 + }, + { + "epoch": 0.24890829694323144, + "grad_norm": 0.5395748717140287, + "learning_rate": 4.141193595342067e-05, + "loss": 0.4052, + "step": 570 + }, + { + "epoch": 0.25109170305676853, + "grad_norm": 0.5834684407168078, + "learning_rate": 4.177583697234352e-05, + "loss": 0.3931, + "step": 575 + }, + { + "epoch": 0.25327510917030566, + "grad_norm": 0.6149399366377413, + "learning_rate": 4.2139737991266375e-05, + "loss": 0.4199, + "step": 580 + }, + { + "epoch": 0.2554585152838428, + "grad_norm": 0.6439187925066173, + "learning_rate": 4.250363901018923e-05, + "loss": 0.3817, + "step": 585 + }, + { + "epoch": 0.2576419213973799, + "grad_norm": 0.7335269690355521, + "learning_rate": 4.286754002911209e-05, + "loss": 0.4146, + "step": 590 + }, + { + "epoch": 0.259825327510917, + "grad_norm": 0.5948480037959886, + "learning_rate": 4.323144104803494e-05, + "loss": 0.3894, + "step": 595 + }, + { + "epoch": 0.26200873362445415, + "grad_norm": 0.7265886088550407, + "learning_rate": 4.359534206695779e-05, + "loss": 0.4099, + "step": 600 + }, + { + "epoch": 0.26419213973799127, + "grad_norm": 0.8146189489413153, + "learning_rate": 4.395924308588064e-05, + "loss": 0.3937, + "step": 605 + }, + { + "epoch": 0.2663755458515284, + "grad_norm": 0.7479919188399701, + "learning_rate": 4.432314410480349e-05, + "loss": 0.4041, + "step": 610 + }, + { + "epoch": 0.2685589519650655, + "grad_norm": 0.6451793400170305, + "learning_rate": 4.468704512372635e-05, + "loss": 0.3923, + "step": 615 + }, + { + "epoch": 0.27074235807860264, + "grad_norm": 0.6731070339864949, + "learning_rate": 4.50509461426492e-05, + "loss": 0.3933, + "step": 620 + }, + { + "epoch": 0.27292576419213976, + "grad_norm": 0.7633021827753019, + "learning_rate": 4.5414847161572056e-05, + "loss": 0.3887, + "step": 625 + }, + { + "epoch": 0.27510917030567683, + "grad_norm": 0.5318275158910994, + "learning_rate": 4.577874818049491e-05, + "loss": 0.3767, + "step": 630 + }, + { + "epoch": 0.27729257641921395, + "grad_norm": 0.6053347759447276, + "learning_rate": 4.614264919941776e-05, + "loss": 0.3996, + "step": 635 + }, + { + "epoch": 0.2794759825327511, + "grad_norm": 0.8239318519157607, + "learning_rate": 4.6506550218340614e-05, + "loss": 0.3966, + "step": 640 + }, + { + "epoch": 0.2816593886462882, + "grad_norm": 0.6239015579696312, + "learning_rate": 4.687045123726347e-05, + "loss": 0.4006, + "step": 645 + }, + { + "epoch": 0.2838427947598253, + "grad_norm": 0.5211608695178815, + "learning_rate": 4.723435225618632e-05, + "loss": 0.3923, + "step": 650 + }, + { + "epoch": 0.28602620087336245, + "grad_norm": 0.6758177772158337, + "learning_rate": 4.759825327510917e-05, + "loss": 0.402, + "step": 655 + }, + { + "epoch": 0.28820960698689957, + "grad_norm": 0.7053141851010356, + "learning_rate": 4.7962154294032025e-05, + "loss": 0.4099, + "step": 660 + }, + { + "epoch": 0.2903930131004367, + "grad_norm": 0.6398424235305709, + "learning_rate": 4.832605531295488e-05, + "loss": 0.3886, + "step": 665 + }, + { + "epoch": 0.2925764192139738, + "grad_norm": 0.6225872999277295, + "learning_rate": 4.868995633187774e-05, + "loss": 0.3706, + "step": 670 + }, + { + "epoch": 0.29475982532751094, + "grad_norm": 0.5559572514117758, + "learning_rate": 4.905385735080058e-05, + "loss": 0.4012, + "step": 675 + }, + { + "epoch": 0.29694323144104806, + "grad_norm": 0.5432885290294175, + "learning_rate": 4.9417758369723435e-05, + "loss": 0.3842, + "step": 680 + }, + { + "epoch": 0.29912663755458513, + "grad_norm": 0.6340294335348677, + "learning_rate": 4.978165938864629e-05, + "loss": 0.3855, + "step": 685 + }, + { + "epoch": 0.30131004366812225, + "grad_norm": 0.7186273990805551, + "learning_rate": 4.999998709163553e-05, + "loss": 0.3705, + "step": 690 + }, + { + "epoch": 0.3034934497816594, + "grad_norm": 0.6162542024063481, + "learning_rate": 4.999984187268831e-05, + "loss": 0.3828, + "step": 695 + }, + { + "epoch": 0.3056768558951965, + "grad_norm": 0.588350309953633, + "learning_rate": 4.9999535300278666e-05, + "loss": 0.3763, + "step": 700 + }, + { + "epoch": 0.3078602620087336, + "grad_norm": 0.5721216258057977, + "learning_rate": 4.999906737638528e-05, + "loss": 0.3928, + "step": 705 + }, + { + "epoch": 0.31004366812227074, + "grad_norm": 1.0000585963765807, + "learning_rate": 4.9998438104028226e-05, + "loss": 0.3926, + "step": 710 + }, + { + "epoch": 0.31222707423580787, + "grad_norm": 2.2978661693892355, + "learning_rate": 4.999764748726891e-05, + "loss": 0.385, + "step": 715 + }, + { + "epoch": 0.314410480349345, + "grad_norm": 1.8144056170150589, + "learning_rate": 4.9996695531210135e-05, + "loss": 0.4175, + "step": 720 + }, + { + "epoch": 0.3165938864628821, + "grad_norm": 0.7618077273846784, + "learning_rate": 4.9995582241996e-05, + "loss": 0.4076, + "step": 725 + }, + { + "epoch": 0.31877729257641924, + "grad_norm": 0.8570756785316501, + "learning_rate": 4.9994307626811865e-05, + "loss": 0.3912, + "step": 730 + }, + { + "epoch": 0.32096069868995636, + "grad_norm": 0.6575118539772534, + "learning_rate": 4.999287169388433e-05, + "loss": 0.3913, + "step": 735 + }, + { + "epoch": 0.3231441048034934, + "grad_norm": 0.7341799315850096, + "learning_rate": 4.9991274452481164e-05, + "loss": 0.4078, + "step": 740 + }, + { + "epoch": 0.32532751091703055, + "grad_norm": 0.6776721860026673, + "learning_rate": 4.9989515912911234e-05, + "loss": 0.3957, + "step": 745 + }, + { + "epoch": 0.32751091703056767, + "grad_norm": 0.6298219471990844, + "learning_rate": 4.998759608652449e-05, + "loss": 0.4049, + "step": 750 + }, + { + "epoch": 0.3296943231441048, + "grad_norm": 0.5126548037982753, + "learning_rate": 4.998551498571182e-05, + "loss": 0.3929, + "step": 755 + }, + { + "epoch": 0.3318777292576419, + "grad_norm": 0.6814430013566892, + "learning_rate": 4.9983272623905045e-05, + "loss": 0.3823, + "step": 760 + }, + { + "epoch": 0.33406113537117904, + "grad_norm": 0.5643156159588145, + "learning_rate": 4.9980869015576736e-05, + "loss": 0.3761, + "step": 765 + }, + { + "epoch": 0.33624454148471616, + "grad_norm": 0.4499045190804366, + "learning_rate": 4.997830417624024e-05, + "loss": 0.3867, + "step": 770 + }, + { + "epoch": 0.3384279475982533, + "grad_norm": 0.7955611854097291, + "learning_rate": 4.9975578122449485e-05, + "loss": 0.3953, + "step": 775 + }, + { + "epoch": 0.3406113537117904, + "grad_norm": 0.5150140528502942, + "learning_rate": 4.99726908717989e-05, + "loss": 0.3769, + "step": 780 + }, + { + "epoch": 0.34279475982532753, + "grad_norm": 0.5896604319461464, + "learning_rate": 4.9969642442923325e-05, + "loss": 0.3776, + "step": 785 + }, + { + "epoch": 0.34497816593886466, + "grad_norm": 0.7333319728063213, + "learning_rate": 4.996643285549787e-05, + "loss": 0.3912, + "step": 790 + }, + { + "epoch": 0.3471615720524017, + "grad_norm": 0.975012641228752, + "learning_rate": 4.996306213023778e-05, + "loss": 0.374, + "step": 795 + }, + { + "epoch": 0.34934497816593885, + "grad_norm": 0.6363914305855141, + "learning_rate": 4.995953028889833e-05, + "loss": 0.3908, + "step": 800 + }, + { + "epoch": 0.35152838427947597, + "grad_norm": 0.7256711049149246, + "learning_rate": 4.995583735427465e-05, + "loss": 0.3804, + "step": 805 + }, + { + "epoch": 0.3537117903930131, + "grad_norm": 0.5201509210204703, + "learning_rate": 4.9951983350201604e-05, + "loss": 0.3787, + "step": 810 + }, + { + "epoch": 0.3558951965065502, + "grad_norm": 0.5937818267570265, + "learning_rate": 4.9947968301553625e-05, + "loss": 0.3681, + "step": 815 + }, + { + "epoch": 0.35807860262008734, + "grad_norm": 0.5805445788983403, + "learning_rate": 4.994379223424456e-05, + "loss": 0.3632, + "step": 820 + }, + { + "epoch": 0.36026200873362446, + "grad_norm": 0.4804762583689749, + "learning_rate": 4.99394551752275e-05, + "loss": 0.3729, + "step": 825 + }, + { + "epoch": 0.3624454148471616, + "grad_norm": 0.4780935144472406, + "learning_rate": 4.9934957152494585e-05, + "loss": 0.366, + "step": 830 + }, + { + "epoch": 0.3646288209606987, + "grad_norm": 0.46427385625879564, + "learning_rate": 4.993029819507687e-05, + "loss": 0.3641, + "step": 835 + }, + { + "epoch": 0.36681222707423583, + "grad_norm": 0.5283252072301436, + "learning_rate": 4.9925478333044114e-05, + "loss": 0.3797, + "step": 840 + }, + { + "epoch": 0.36899563318777295, + "grad_norm": 0.5593669993363974, + "learning_rate": 4.992049759750455e-05, + "loss": 0.3776, + "step": 845 + }, + { + "epoch": 0.37117903930131, + "grad_norm": 0.6282087870379117, + "learning_rate": 4.9915356020604754e-05, + "loss": 0.3723, + "step": 850 + }, + { + "epoch": 0.37336244541484714, + "grad_norm": 0.4876112040234335, + "learning_rate": 4.9910053635529376e-05, + "loss": 0.3918, + "step": 855 + }, + { + "epoch": 0.37554585152838427, + "grad_norm": 0.4642413234429665, + "learning_rate": 4.9904590476500955e-05, + "loss": 0.3809, + "step": 860 + }, + { + "epoch": 0.3777292576419214, + "grad_norm": 0.44822878442026826, + "learning_rate": 4.989896657877972e-05, + "loss": 0.37, + "step": 865 + }, + { + "epoch": 0.3799126637554585, + "grad_norm": 0.4846478506665109, + "learning_rate": 4.98931819786633e-05, + "loss": 0.3742, + "step": 870 + }, + { + "epoch": 0.38209606986899564, + "grad_norm": 0.4166354564636158, + "learning_rate": 4.988723671348654e-05, + "loss": 0.3658, + "step": 875 + }, + { + "epoch": 0.38427947598253276, + "grad_norm": 0.49223225972882195, + "learning_rate": 4.9881130821621255e-05, + "loss": 0.3914, + "step": 880 + }, + { + "epoch": 0.3864628820960699, + "grad_norm": 0.526884770677872, + "learning_rate": 4.987486434247596e-05, + "loss": 0.3754, + "step": 885 + }, + { + "epoch": 0.388646288209607, + "grad_norm": 0.5232997418172833, + "learning_rate": 4.9868437316495636e-05, + "loss": 0.3724, + "step": 890 + }, + { + "epoch": 0.39082969432314413, + "grad_norm": 0.48376676230007215, + "learning_rate": 4.9861849785161465e-05, + "loss": 0.3673, + "step": 895 + }, + { + "epoch": 0.3930131004366812, + "grad_norm": 0.5294590310273921, + "learning_rate": 4.9855101790990556e-05, + "loss": 0.3727, + "step": 900 + }, + { + "epoch": 0.3951965065502183, + "grad_norm": 0.47896380765481217, + "learning_rate": 4.9848193377535665e-05, + "loss": 0.3841, + "step": 905 + }, + { + "epoch": 0.39737991266375544, + "grad_norm": 0.607838261659882, + "learning_rate": 4.9841124589384944e-05, + "loss": 0.3723, + "step": 910 + }, + { + "epoch": 0.39956331877729256, + "grad_norm": 0.5096948643379262, + "learning_rate": 4.983389547216161e-05, + "loss": 0.3707, + "step": 915 + }, + { + "epoch": 0.4017467248908297, + "grad_norm": 0.38304781823406214, + "learning_rate": 4.9826506072523684e-05, + "loss": 0.3545, + "step": 920 + }, + { + "epoch": 0.4039301310043668, + "grad_norm": 0.587506061988026, + "learning_rate": 4.981895643816367e-05, + "loss": 0.3718, + "step": 925 + }, + { + "epoch": 0.40611353711790393, + "grad_norm": 0.5967719221364212, + "learning_rate": 4.9811246617808275e-05, + "loss": 0.3684, + "step": 930 + }, + { + "epoch": 0.40829694323144106, + "grad_norm": 0.6935599142313973, + "learning_rate": 4.980337666121806e-05, + "loss": 0.3789, + "step": 935 + }, + { + "epoch": 0.4104803493449782, + "grad_norm": 0.5766117112149044, + "learning_rate": 4.979534661918712e-05, + "loss": 0.3542, + "step": 940 + }, + { + "epoch": 0.4126637554585153, + "grad_norm": 0.43795584226862594, + "learning_rate": 4.978715654354281e-05, + "loss": 0.3595, + "step": 945 + }, + { + "epoch": 0.4148471615720524, + "grad_norm": 0.4430509318673314, + "learning_rate": 4.9778806487145326e-05, + "loss": 0.3666, + "step": 950 + }, + { + "epoch": 0.4170305676855895, + "grad_norm": 0.4481773430888534, + "learning_rate": 4.977029650388744e-05, + "loss": 0.3654, + "step": 955 + }, + { + "epoch": 0.4192139737991266, + "grad_norm": 0.4657407424462558, + "learning_rate": 4.976162664869411e-05, + "loss": 0.3542, + "step": 960 + }, + { + "epoch": 0.42139737991266374, + "grad_norm": 0.5287646283190662, + "learning_rate": 4.975279697752213e-05, + "loss": 0.3594, + "step": 965 + }, + { + "epoch": 0.42358078602620086, + "grad_norm": 0.4887455966923865, + "learning_rate": 4.974380754735979e-05, + "loss": 0.3669, + "step": 970 + }, + { + "epoch": 0.425764192139738, + "grad_norm": 0.43872446154490324, + "learning_rate": 4.9734658416226476e-05, + "loss": 0.3616, + "step": 975 + }, + { + "epoch": 0.4279475982532751, + "grad_norm": 0.6424735520245978, + "learning_rate": 4.972534964317233e-05, + "loss": 0.369, + "step": 980 + }, + { + "epoch": 0.43013100436681223, + "grad_norm": 0.48958723865804354, + "learning_rate": 4.9715881288277835e-05, + "loss": 0.3594, + "step": 985 + }, + { + "epoch": 0.43231441048034935, + "grad_norm": 0.4479761879052664, + "learning_rate": 4.9706253412653446e-05, + "loss": 0.3549, + "step": 990 + }, + { + "epoch": 0.4344978165938865, + "grad_norm": 0.5435346773895244, + "learning_rate": 4.9696466078439216e-05, + "loss": 0.3671, + "step": 995 + }, + { + "epoch": 0.4366812227074236, + "grad_norm": 0.5308581082448025, + "learning_rate": 4.968651934880434e-05, + "loss": 0.365, + "step": 1000 + }, + { + "epoch": 0.4388646288209607, + "grad_norm": 0.39896740398055186, + "learning_rate": 4.96764132879468e-05, + "loss": 0.3598, + "step": 1005 + }, + { + "epoch": 0.4410480349344978, + "grad_norm": 0.5067575654052805, + "learning_rate": 4.9666147961092924e-05, + "loss": 0.3659, + "step": 1010 + }, + { + "epoch": 0.4432314410480349, + "grad_norm": 0.5592601406097657, + "learning_rate": 4.9655723434496984e-05, + "loss": 0.3719, + "step": 1015 + }, + { + "epoch": 0.44541484716157204, + "grad_norm": 0.4413849291952618, + "learning_rate": 4.9645139775440716e-05, + "loss": 0.3652, + "step": 1020 + }, + { + "epoch": 0.44759825327510916, + "grad_norm": 0.3910244442044385, + "learning_rate": 4.9634397052232983e-05, + "loss": 0.3691, + "step": 1025 + }, + { + "epoch": 0.4497816593886463, + "grad_norm": 0.46071657374615793, + "learning_rate": 4.9623495334209226e-05, + "loss": 0.3624, + "step": 1030 + }, + { + "epoch": 0.4519650655021834, + "grad_norm": 0.4226092000790114, + "learning_rate": 4.96124346917311e-05, + "loss": 0.3676, + "step": 1035 + }, + { + "epoch": 0.45414847161572053, + "grad_norm": 0.44419459860035787, + "learning_rate": 4.960121519618597e-05, + "loss": 0.332, + "step": 1040 + }, + { + "epoch": 0.45633187772925765, + "grad_norm": 0.3577954443167031, + "learning_rate": 4.9589836919986475e-05, + "loss": 0.3512, + "step": 1045 + }, + { + "epoch": 0.4585152838427948, + "grad_norm": 0.5820450329129828, + "learning_rate": 4.957829993657005e-05, + "loss": 0.3501, + "step": 1050 + }, + { + "epoch": 0.4606986899563319, + "grad_norm": 0.5150191150181325, + "learning_rate": 4.956660432039846e-05, + "loss": 0.3597, + "step": 1055 + }, + { + "epoch": 0.462882096069869, + "grad_norm": 0.4319371225817441, + "learning_rate": 4.955475014695729e-05, + "loss": 0.3676, + "step": 1060 + }, + { + "epoch": 0.4650655021834061, + "grad_norm": 0.5533636569254564, + "learning_rate": 4.954273749275552e-05, + "loss": 0.3727, + "step": 1065 + }, + { + "epoch": 0.4672489082969432, + "grad_norm": 0.4958192610268456, + "learning_rate": 4.9530566435324963e-05, + "loss": 0.3488, + "step": 1070 + }, + { + "epoch": 0.46943231441048033, + "grad_norm": 0.5108202025863451, + "learning_rate": 4.9518237053219816e-05, + "loss": 0.3709, + "step": 1075 + }, + { + "epoch": 0.47161572052401746, + "grad_norm": 0.45372615060906546, + "learning_rate": 4.950574942601611e-05, + "loss": 0.3475, + "step": 1080 + }, + { + "epoch": 0.4737991266375546, + "grad_norm": 0.43181104589299185, + "learning_rate": 4.9493103634311245e-05, + "loss": 0.3527, + "step": 1085 + }, + { + "epoch": 0.4759825327510917, + "grad_norm": 0.49622919369152685, + "learning_rate": 4.948029975972341e-05, + "loss": 0.34, + "step": 1090 + }, + { + "epoch": 0.4781659388646288, + "grad_norm": 0.5352955894735107, + "learning_rate": 4.9467337884891127e-05, + "loss": 0.3578, + "step": 1095 + }, + { + "epoch": 0.48034934497816595, + "grad_norm": 0.48767956685977704, + "learning_rate": 4.945421809347266e-05, + "loss": 0.3437, + "step": 1100 + }, + { + "epoch": 0.48253275109170307, + "grad_norm": 0.5172947775347728, + "learning_rate": 4.944094047014547e-05, + "loss": 0.3582, + "step": 1105 + }, + { + "epoch": 0.4847161572052402, + "grad_norm": 0.4633995942526625, + "learning_rate": 4.942750510060574e-05, + "loss": 0.3369, + "step": 1110 + }, + { + "epoch": 0.4868995633187773, + "grad_norm": 0.5384683843923247, + "learning_rate": 4.9413912071567756e-05, + "loss": 0.3683, + "step": 1115 + }, + { + "epoch": 0.4890829694323144, + "grad_norm": 0.428948742370993, + "learning_rate": 4.940016147076336e-05, + "loss": 0.3425, + "step": 1120 + }, + { + "epoch": 0.4912663755458515, + "grad_norm": 0.5160388037360805, + "learning_rate": 4.938625338694139e-05, + "loss": 0.3609, + "step": 1125 + }, + { + "epoch": 0.49344978165938863, + "grad_norm": 0.45537160535635335, + "learning_rate": 4.937218790986712e-05, + "loss": 0.3532, + "step": 1130 + }, + { + "epoch": 0.49563318777292575, + "grad_norm": 0.4996736665030406, + "learning_rate": 4.935796513032165e-05, + "loss": 0.3434, + "step": 1135 + }, + { + "epoch": 0.4978165938864629, + "grad_norm": 0.5155894229667807, + "learning_rate": 4.934358514010137e-05, + "loss": 0.3569, + "step": 1140 + }, + { + "epoch": 0.5, + "grad_norm": 0.38793250116768685, + "learning_rate": 4.93290480320173e-05, + "loss": 0.3545, + "step": 1145 + }, + { + "epoch": 0.5021834061135371, + "grad_norm": 0.36522790796535515, + "learning_rate": 4.931435389989455e-05, + "loss": 0.3482, + "step": 1150 + }, + { + "epoch": 0.5043668122270742, + "grad_norm": 0.46762083247492375, + "learning_rate": 4.929950283857167e-05, + "loss": 0.3704, + "step": 1155 + }, + { + "epoch": 0.5065502183406113, + "grad_norm": 0.45795606465334227, + "learning_rate": 4.9284494943900085e-05, + "loss": 0.3646, + "step": 1160 + }, + { + "epoch": 0.5087336244541485, + "grad_norm": 0.42995508768886065, + "learning_rate": 4.926933031274344e-05, + "loss": 0.3626, + "step": 1165 + }, + { + "epoch": 0.5109170305676856, + "grad_norm": 0.4224390944477537, + "learning_rate": 4.925400904297698e-05, + "loss": 0.347, + "step": 1170 + }, + { + "epoch": 0.5131004366812227, + "grad_norm": 0.5046566006479356, + "learning_rate": 4.923853123348693e-05, + "loss": 0.3678, + "step": 1175 + }, + { + "epoch": 0.5152838427947598, + "grad_norm": 0.5310348130320324, + "learning_rate": 4.922289698416983e-05, + "loss": 0.3568, + "step": 1180 + }, + { + "epoch": 0.517467248908297, + "grad_norm": 0.44019956240689734, + "learning_rate": 4.920710639593196e-05, + "loss": 0.3555, + "step": 1185 + }, + { + "epoch": 0.519650655021834, + "grad_norm": 0.36638621076470246, + "learning_rate": 4.919115957068859e-05, + "loss": 0.3529, + "step": 1190 + }, + { + "epoch": 0.5218340611353712, + "grad_norm": 0.4899606255024857, + "learning_rate": 4.917505661136339e-05, + "loss": 0.3491, + "step": 1195 + }, + { + "epoch": 0.5240174672489083, + "grad_norm": 0.501601743259068, + "learning_rate": 4.915879762188774e-05, + "loss": 0.3615, + "step": 1200 + }, + { + "epoch": 0.5262008733624454, + "grad_norm": 0.6025124136368822, + "learning_rate": 4.9142382707200096e-05, + "loss": 0.3553, + "step": 1205 + }, + { + "epoch": 0.5283842794759825, + "grad_norm": 0.445329859368048, + "learning_rate": 4.912581197324524e-05, + "loss": 0.3633, + "step": 1210 + }, + { + "epoch": 0.5305676855895196, + "grad_norm": 0.6418024421311272, + "learning_rate": 4.9109085526973674e-05, + "loss": 0.3476, + "step": 1215 + }, + { + "epoch": 0.5327510917030568, + "grad_norm": 0.4430605711065509, + "learning_rate": 4.909220347634087e-05, + "loss": 0.359, + "step": 1220 + }, + { + "epoch": 0.5349344978165939, + "grad_norm": 0.4354421223792483, + "learning_rate": 4.907516593030662e-05, + "loss": 0.3515, + "step": 1225 + }, + { + "epoch": 0.537117903930131, + "grad_norm": 0.47281124382089545, + "learning_rate": 4.905797299883429e-05, + "loss": 0.3629, + "step": 1230 + }, + { + "epoch": 0.5393013100436681, + "grad_norm": 0.4468031895316623, + "learning_rate": 4.904062479289015e-05, + "loss": 0.3484, + "step": 1235 + }, + { + "epoch": 0.5414847161572053, + "grad_norm": 0.4659670701752814, + "learning_rate": 4.902312142444264e-05, + "loss": 0.3526, + "step": 1240 + }, + { + "epoch": 0.5436681222707423, + "grad_norm": 0.371869503352742, + "learning_rate": 4.900546300646162e-05, + "loss": 0.3366, + "step": 1245 + }, + { + "epoch": 0.5458515283842795, + "grad_norm": 0.42423288318858376, + "learning_rate": 4.89876496529177e-05, + "loss": 0.3543, + "step": 1250 + }, + { + "epoch": 0.5480349344978166, + "grad_norm": 0.4609253407888565, + "learning_rate": 4.896968147878145e-05, + "loss": 0.3565, + "step": 1255 + }, + { + "epoch": 0.5502183406113537, + "grad_norm": 0.3237267708873633, + "learning_rate": 4.895155860002271e-05, + "loss": 0.3377, + "step": 1260 + }, + { + "epoch": 0.5524017467248908, + "grad_norm": 0.4546639718401833, + "learning_rate": 4.893328113360975e-05, + "loss": 0.3535, + "step": 1265 + }, + { + "epoch": 0.5545851528384279, + "grad_norm": 0.43148670011250856, + "learning_rate": 4.891484919750865e-05, + "loss": 0.344, + "step": 1270 + }, + { + "epoch": 0.5567685589519651, + "grad_norm": 0.3991105241595635, + "learning_rate": 4.889626291068241e-05, + "loss": 0.339, + "step": 1275 + }, + { + "epoch": 0.5589519650655022, + "grad_norm": 0.48987054857422774, + "learning_rate": 4.887752239309027e-05, + "loss": 0.3391, + "step": 1280 + }, + { + "epoch": 0.5611353711790393, + "grad_norm": 0.4609741215989259, + "learning_rate": 4.885862776568689e-05, + "loss": 0.3576, + "step": 1285 + }, + { + "epoch": 0.5633187772925764, + "grad_norm": 0.50704867138633, + "learning_rate": 4.883957915042158e-05, + "loss": 0.3496, + "step": 1290 + }, + { + "epoch": 0.5655021834061136, + "grad_norm": 0.48159077762660746, + "learning_rate": 4.882037667023752e-05, + "loss": 0.3361, + "step": 1295 + }, + { + "epoch": 0.5676855895196506, + "grad_norm": 0.7062981902269905, + "learning_rate": 4.880102044907096e-05, + "loss": 0.3677, + "step": 1300 + }, + { + "epoch": 0.5698689956331878, + "grad_norm": 0.4888521580220186, + "learning_rate": 4.878151061185042e-05, + "loss": 0.3294, + "step": 1305 + }, + { + "epoch": 0.5720524017467249, + "grad_norm": 0.4540470014267252, + "learning_rate": 4.876184728449589e-05, + "loss": 0.3488, + "step": 1310 + }, + { + "epoch": 0.574235807860262, + "grad_norm": 0.3930373040922702, + "learning_rate": 4.874203059391802e-05, + "loss": 0.3364, + "step": 1315 + }, + { + "epoch": 0.5764192139737991, + "grad_norm": 0.4451157942220829, + "learning_rate": 4.8722060668017267e-05, + "loss": 0.3562, + "step": 1320 + }, + { + "epoch": 0.5786026200873362, + "grad_norm": 0.3942536599077881, + "learning_rate": 4.870193763568312e-05, + "loss": 0.3619, + "step": 1325 + }, + { + "epoch": 0.5807860262008734, + "grad_norm": 0.48861179907906854, + "learning_rate": 4.8681661626793246e-05, + "loss": 0.3468, + "step": 1330 + }, + { + "epoch": 0.5829694323144105, + "grad_norm": 0.42138612451614066, + "learning_rate": 4.866123277221264e-05, + "loss": 0.347, + "step": 1335 + }, + { + "epoch": 0.5851528384279476, + "grad_norm": 0.4074290406551753, + "learning_rate": 4.864065120379278e-05, + "loss": 0.3573, + "step": 1340 + }, + { + "epoch": 0.5873362445414847, + "grad_norm": 0.3987198118209572, + "learning_rate": 4.861991705437081e-05, + "loss": 0.3598, + "step": 1345 + }, + { + "epoch": 0.5895196506550219, + "grad_norm": 0.46460108590367194, + "learning_rate": 4.859903045776866e-05, + "loss": 0.3524, + "step": 1350 + }, + { + "epoch": 0.5917030567685589, + "grad_norm": 0.4803498793618286, + "learning_rate": 4.857799154879215e-05, + "loss": 0.3605, + "step": 1355 + }, + { + "epoch": 0.5938864628820961, + "grad_norm": 0.5901063012029109, + "learning_rate": 4.8556800463230165e-05, + "loss": 0.3437, + "step": 1360 + }, + { + "epoch": 0.5960698689956332, + "grad_norm": 0.4641267585126181, + "learning_rate": 4.8535457337853794e-05, + "loss": 0.3285, + "step": 1365 + }, + { + "epoch": 0.5982532751091703, + "grad_norm": 0.3647796263943895, + "learning_rate": 4.8513962310415396e-05, + "loss": 0.3463, + "step": 1370 + }, + { + "epoch": 0.6004366812227074, + "grad_norm": 0.3648828726715139, + "learning_rate": 4.849231551964771e-05, + "loss": 0.3535, + "step": 1375 + }, + { + "epoch": 0.6026200873362445, + "grad_norm": 0.40542343011346316, + "learning_rate": 4.847051710526302e-05, + "loss": 0.3437, + "step": 1380 + }, + { + "epoch": 0.6048034934497817, + "grad_norm": 0.350582105718404, + "learning_rate": 4.8448567207952205e-05, + "loss": 0.3549, + "step": 1385 + }, + { + "epoch": 0.6069868995633187, + "grad_norm": 0.39947990507599496, + "learning_rate": 4.8426465969383826e-05, + "loss": 0.3243, + "step": 1390 + }, + { + "epoch": 0.6091703056768559, + "grad_norm": 0.4129694646452404, + "learning_rate": 4.840421353220326e-05, + "loss": 0.3251, + "step": 1395 + }, + { + "epoch": 0.611353711790393, + "grad_norm": 0.3820909710575323, + "learning_rate": 4.83818100400317e-05, + "loss": 0.334, + "step": 1400 + }, + { + "epoch": 0.6135371179039302, + "grad_norm": 0.4069293886007247, + "learning_rate": 4.835925563746532e-05, + "loss": 0.3551, + "step": 1405 + }, + { + "epoch": 0.6157205240174672, + "grad_norm": 0.4582430996524103, + "learning_rate": 4.833655047007427e-05, + "loss": 0.3504, + "step": 1410 + }, + { + "epoch": 0.6179039301310044, + "grad_norm": 0.3828069417511488, + "learning_rate": 4.831369468440177e-05, + "loss": 0.3358, + "step": 1415 + }, + { + "epoch": 0.6200873362445415, + "grad_norm": 0.379141828052868, + "learning_rate": 4.8290688427963175e-05, + "loss": 0.3315, + "step": 1420 + }, + { + "epoch": 0.6222707423580786, + "grad_norm": 0.4059377645684439, + "learning_rate": 4.8267531849244974e-05, + "loss": 0.3403, + "step": 1425 + }, + { + "epoch": 0.6244541484716157, + "grad_norm": 0.3781489511039065, + "learning_rate": 4.8244225097703885e-05, + "loss": 0.351, + "step": 1430 + }, + { + "epoch": 0.6266375545851528, + "grad_norm": 0.36811899180808383, + "learning_rate": 4.8220768323765866e-05, + "loss": 0.345, + "step": 1435 + }, + { + "epoch": 0.62882096069869, + "grad_norm": 0.48695298818921573, + "learning_rate": 4.819716167882513e-05, + "loss": 0.343, + "step": 1440 + }, + { + "epoch": 0.631004366812227, + "grad_norm": 0.41424246411412347, + "learning_rate": 4.8173405315243216e-05, + "loss": 0.3381, + "step": 1445 + }, + { + "epoch": 0.6331877729257642, + "grad_norm": 0.5982905072375967, + "learning_rate": 4.814949938634793e-05, + "loss": 0.3334, + "step": 1450 + }, + { + "epoch": 0.6353711790393013, + "grad_norm": 0.41432212938369534, + "learning_rate": 4.8125444046432434e-05, + "loss": 0.3387, + "step": 1455 + }, + { + "epoch": 0.6375545851528385, + "grad_norm": 0.3970797596419897, + "learning_rate": 4.810123945075421e-05, + "loss": 0.3485, + "step": 1460 + }, + { + "epoch": 0.6397379912663755, + "grad_norm": 0.36649659825973896, + "learning_rate": 4.807688575553406e-05, + "loss": 0.3381, + "step": 1465 + }, + { + "epoch": 0.6419213973799127, + "grad_norm": 0.4151615051412959, + "learning_rate": 4.80523831179551e-05, + "loss": 0.3316, + "step": 1470 + }, + { + "epoch": 0.6441048034934498, + "grad_norm": 0.41296409849005844, + "learning_rate": 4.8027731696161735e-05, + "loss": 0.3378, + "step": 1475 + }, + { + "epoch": 0.6462882096069869, + "grad_norm": 0.43599613227274564, + "learning_rate": 4.8002931649258674e-05, + "loss": 0.3343, + "step": 1480 + }, + { + "epoch": 0.648471615720524, + "grad_norm": 0.5041455481985673, + "learning_rate": 4.7977983137309855e-05, + "loss": 0.3337, + "step": 1485 + }, + { + "epoch": 0.6506550218340611, + "grad_norm": 0.41603606102503005, + "learning_rate": 4.795288632133747e-05, + "loss": 0.3328, + "step": 1490 + }, + { + "epoch": 0.6528384279475983, + "grad_norm": 0.4091756291458897, + "learning_rate": 4.792764136332084e-05, + "loss": 0.3558, + "step": 1495 + }, + { + "epoch": 0.6550218340611353, + "grad_norm": 0.3535456736539121, + "learning_rate": 4.790224842619548e-05, + "loss": 0.3305, + "step": 1500 + }, + { + "epoch": 0.6572052401746725, + "grad_norm": 0.4204136805374324, + "learning_rate": 4.787670767385194e-05, + "loss": 0.3399, + "step": 1505 + }, + { + "epoch": 0.6593886462882096, + "grad_norm": 0.4138703097114033, + "learning_rate": 4.785101927113482e-05, + "loss": 0.3423, + "step": 1510 + }, + { + "epoch": 0.6615720524017468, + "grad_norm": 0.42867964938598574, + "learning_rate": 4.782518338384168e-05, + "loss": 0.3474, + "step": 1515 + }, + { + "epoch": 0.6637554585152838, + "grad_norm": 0.3655864787630958, + "learning_rate": 4.779920017872197e-05, + "loss": 0.3426, + "step": 1520 + }, + { + "epoch": 0.665938864628821, + "grad_norm": 0.31312668856989573, + "learning_rate": 4.7773069823475945e-05, + "loss": 0.3317, + "step": 1525 + }, + { + "epoch": 0.6681222707423581, + "grad_norm": 0.3720008834724695, + "learning_rate": 4.774679248675362e-05, + "loss": 0.3468, + "step": 1530 + }, + { + "epoch": 0.6703056768558951, + "grad_norm": 0.3500906347775602, + "learning_rate": 4.772036833815363e-05, + "loss": 0.3305, + "step": 1535 + }, + { + "epoch": 0.6724890829694323, + "grad_norm": 0.4324016765021572, + "learning_rate": 4.7693797548222166e-05, + "loss": 0.3226, + "step": 1540 + }, + { + "epoch": 0.6746724890829694, + "grad_norm": 0.40696438387223544, + "learning_rate": 4.766708028845188e-05, + "loss": 0.3325, + "step": 1545 + }, + { + "epoch": 0.6768558951965066, + "grad_norm": 0.4155361411655341, + "learning_rate": 4.764021673128075e-05, + "loss": 0.3482, + "step": 1550 + }, + { + "epoch": 0.6790393013100436, + "grad_norm": 0.4890226077260248, + "learning_rate": 4.761320705009099e-05, + "loss": 0.3457, + "step": 1555 + }, + { + "epoch": 0.6812227074235808, + "grad_norm": 0.42425030996602786, + "learning_rate": 4.758605141920792e-05, + "loss": 0.3512, + "step": 1560 + }, + { + "epoch": 0.6834061135371179, + "grad_norm": 0.3594877731084044, + "learning_rate": 4.755875001389887e-05, + "loss": 0.3434, + "step": 1565 + }, + { + "epoch": 0.6855895196506551, + "grad_norm": 0.3979846319153574, + "learning_rate": 4.753130301037199e-05, + "loss": 0.3337, + "step": 1570 + }, + { + "epoch": 0.6877729257641921, + "grad_norm": 0.402834791445396, + "learning_rate": 4.7503710585775174e-05, + "loss": 0.3387, + "step": 1575 + }, + { + "epoch": 0.6899563318777293, + "grad_norm": 0.34220820957249626, + "learning_rate": 4.747597291819487e-05, + "loss": 0.3439, + "step": 1580 + }, + { + "epoch": 0.6921397379912664, + "grad_norm": 0.3831695794480096, + "learning_rate": 4.744809018665495e-05, + "loss": 0.3373, + "step": 1585 + }, + { + "epoch": 0.6943231441048034, + "grad_norm": 0.40957271083541796, + "learning_rate": 4.742006257111558e-05, + "loss": 0.3358, + "step": 1590 + }, + { + "epoch": 0.6965065502183406, + "grad_norm": 0.3989291887534643, + "learning_rate": 4.7391890252472016e-05, + "loss": 0.3624, + "step": 1595 + }, + { + "epoch": 0.6986899563318777, + "grad_norm": 0.47006740071344755, + "learning_rate": 4.736357341255344e-05, + "loss": 0.3427, + "step": 1600 + }, + { + "epoch": 0.7008733624454149, + "grad_norm": 0.5495070509701956, + "learning_rate": 4.7335112234121824e-05, + "loss": 0.3332, + "step": 1605 + }, + { + "epoch": 0.7030567685589519, + "grad_norm": 0.37491891454833864, + "learning_rate": 4.730650690087072e-05, + "loss": 0.3406, + "step": 1610 + }, + { + "epoch": 0.7052401746724891, + "grad_norm": 0.43036507029194554, + "learning_rate": 4.727775759742408e-05, + "loss": 0.3431, + "step": 1615 + }, + { + "epoch": 0.7074235807860262, + "grad_norm": 0.3820782925001961, + "learning_rate": 4.724886450933505e-05, + "loss": 0.3436, + "step": 1620 + }, + { + "epoch": 0.7096069868995634, + "grad_norm": 0.4860255736060773, + "learning_rate": 4.721982782308481e-05, + "loss": 0.3523, + "step": 1625 + }, + { + "epoch": 0.7117903930131004, + "grad_norm": 0.3695141770304634, + "learning_rate": 4.719064772608134e-05, + "loss": 0.3462, + "step": 1630 + }, + { + "epoch": 0.7139737991266376, + "grad_norm": 0.4003960638271693, + "learning_rate": 4.7161324406658225e-05, + "loss": 0.3286, + "step": 1635 + }, + { + "epoch": 0.7161572052401747, + "grad_norm": 0.3530404563540438, + "learning_rate": 4.713185805407342e-05, + "loss": 0.3236, + "step": 1640 + }, + { + "epoch": 0.7183406113537117, + "grad_norm": 0.584075905055993, + "learning_rate": 4.7102248858508044e-05, + "loss": 0.3544, + "step": 1645 + }, + { + "epoch": 0.7205240174672489, + "grad_norm": 0.46166165713972934, + "learning_rate": 4.7072497011065156e-05, + "loss": 0.3298, + "step": 1650 + }, + { + "epoch": 0.722707423580786, + "grad_norm": 0.3355056455618887, + "learning_rate": 4.704260270376853e-05, + "loss": 0.3264, + "step": 1655 + }, + { + "epoch": 0.7248908296943232, + "grad_norm": 0.39311481965484996, + "learning_rate": 4.7012566129561365e-05, + "loss": 0.3381, + "step": 1660 + }, + { + "epoch": 0.7270742358078602, + "grad_norm": 0.38148657969468514, + "learning_rate": 4.6982387482305104e-05, + "loss": 0.3322, + "step": 1665 + }, + { + "epoch": 0.7292576419213974, + "grad_norm": 0.44427131750278887, + "learning_rate": 4.6952066956778155e-05, + "loss": 0.3268, + "step": 1670 + }, + { + "epoch": 0.7314410480349345, + "grad_norm": 0.3857743094672692, + "learning_rate": 4.692160474867463e-05, + "loss": 0.3324, + "step": 1675 + }, + { + "epoch": 0.7336244541484717, + "grad_norm": 0.3899786180476667, + "learning_rate": 4.6891001054603056e-05, + "loss": 0.3374, + "step": 1680 + }, + { + "epoch": 0.7358078602620087, + "grad_norm": 0.3696166334234957, + "learning_rate": 4.6860256072085196e-05, + "loss": 0.3453, + "step": 1685 + }, + { + "epoch": 0.7379912663755459, + "grad_norm": 0.30167847925632657, + "learning_rate": 4.6829369999554664e-05, + "loss": 0.3297, + "step": 1690 + }, + { + "epoch": 0.740174672489083, + "grad_norm": 0.37202185004193333, + "learning_rate": 4.6798343036355704e-05, + "loss": 0.3058, + "step": 1695 + }, + { + "epoch": 0.74235807860262, + "grad_norm": 0.3965117699988383, + "learning_rate": 4.6767175382741914e-05, + "loss": 0.3367, + "step": 1700 + }, + { + "epoch": 0.7445414847161572, + "grad_norm": 0.37265664937861914, + "learning_rate": 4.673586723987491e-05, + "loss": 0.3504, + "step": 1705 + }, + { + "epoch": 0.7467248908296943, + "grad_norm": 0.42824465747750634, + "learning_rate": 4.670441880982307e-05, + "loss": 0.3323, + "step": 1710 + }, + { + "epoch": 0.7489082969432315, + "grad_norm": 0.3708258588256483, + "learning_rate": 4.667283029556019e-05, + "loss": 0.3288, + "step": 1715 + }, + { + "epoch": 0.7510917030567685, + "grad_norm": 0.4271790614273471, + "learning_rate": 4.664110190096421e-05, + "loss": 0.3436, + "step": 1720 + }, + { + "epoch": 0.7532751091703057, + "grad_norm": 0.37662914132162023, + "learning_rate": 4.660923383081588e-05, + "loss": 0.3401, + "step": 1725 + }, + { + "epoch": 0.7554585152838428, + "grad_norm": 0.3574814773268414, + "learning_rate": 4.657722629079744e-05, + "loss": 0.3359, + "step": 1730 + }, + { + "epoch": 0.75764192139738, + "grad_norm": 0.44811194231414175, + "learning_rate": 4.6545079487491285e-05, + "loss": 0.3466, + "step": 1735 + }, + { + "epoch": 0.759825327510917, + "grad_norm": 0.5433785036597635, + "learning_rate": 4.6512793628378657e-05, + "loss": 0.3348, + "step": 1740 + }, + { + "epoch": 0.7620087336244541, + "grad_norm": 0.424131707206492, + "learning_rate": 4.648036892183828e-05, + "loss": 0.3381, + "step": 1745 + }, + { + "epoch": 0.7641921397379913, + "grad_norm": 0.3485719470067596, + "learning_rate": 4.644780557714502e-05, + "loss": 0.3338, + "step": 1750 + }, + { + "epoch": 0.7663755458515283, + "grad_norm": 0.3595654663623888, + "learning_rate": 4.641510380446854e-05, + "loss": 0.3238, + "step": 1755 + }, + { + "epoch": 0.7685589519650655, + "grad_norm": 0.4262032500390272, + "learning_rate": 4.638226381487195e-05, + "loss": 0.3373, + "step": 1760 + }, + { + "epoch": 0.7707423580786026, + "grad_norm": 0.44078101888220983, + "learning_rate": 4.634928582031043e-05, + "loss": 0.3449, + "step": 1765 + }, + { + "epoch": 0.7729257641921398, + "grad_norm": 0.4054216447358104, + "learning_rate": 4.631617003362988e-05, + "loss": 0.3337, + "step": 1770 + }, + { + "epoch": 0.7751091703056768, + "grad_norm": 0.3848597633812537, + "learning_rate": 4.628291666856551e-05, + "loss": 0.3211, + "step": 1775 + }, + { + "epoch": 0.777292576419214, + "grad_norm": 0.3408957154658452, + "learning_rate": 4.6249525939740514e-05, + "loss": 0.3306, + "step": 1780 + }, + { + "epoch": 0.7794759825327511, + "grad_norm": 0.4050082743023021, + "learning_rate": 4.6215998062664636e-05, + "loss": 0.336, + "step": 1785 + }, + { + "epoch": 0.7816593886462883, + "grad_norm": 0.4303256630650177, + "learning_rate": 4.61823332537328e-05, + "loss": 0.3271, + "step": 1790 + }, + { + "epoch": 0.7838427947598253, + "grad_norm": 0.3879573361229342, + "learning_rate": 4.6148531730223735e-05, + "loss": 0.3309, + "step": 1795 + }, + { + "epoch": 0.7860262008733624, + "grad_norm": 0.3340993854538669, + "learning_rate": 4.611459371029853e-05, + "loss": 0.3273, + "step": 1800 + }, + { + "epoch": 0.7882096069868996, + "grad_norm": 0.41220645591308713, + "learning_rate": 4.6080519412999244e-05, + "loss": 0.3256, + "step": 1805 + }, + { + "epoch": 0.7903930131004366, + "grad_norm": 0.4607924969239619, + "learning_rate": 4.604630905824751e-05, + "loss": 0.3344, + "step": 1810 + }, + { + "epoch": 0.7925764192139738, + "grad_norm": 0.3986158317545471, + "learning_rate": 4.601196286684309e-05, + "loss": 0.3301, + "step": 1815 + }, + { + "epoch": 0.7947598253275109, + "grad_norm": 0.31961695415950014, + "learning_rate": 4.597748106046247e-05, + "loss": 0.3305, + "step": 1820 + }, + { + "epoch": 0.7969432314410481, + "grad_norm": 0.390531096769233, + "learning_rate": 4.5942863861657394e-05, + "loss": 0.3359, + "step": 1825 + }, + { + "epoch": 0.7991266375545851, + "grad_norm": 0.3431703506154723, + "learning_rate": 4.590811149385348e-05, + "loss": 0.3422, + "step": 1830 + }, + { + "epoch": 0.8013100436681223, + "grad_norm": 0.41192305785453975, + "learning_rate": 4.5873224181348736e-05, + "loss": 0.3247, + "step": 1835 + }, + { + "epoch": 0.8034934497816594, + "grad_norm": 0.2924629037593409, + "learning_rate": 4.5838202149312145e-05, + "loss": 0.3304, + "step": 1840 + }, + { + "epoch": 0.8056768558951966, + "grad_norm": 0.5634336476609664, + "learning_rate": 4.580304562378218e-05, + "loss": 0.3303, + "step": 1845 + }, + { + "epoch": 0.8078602620087336, + "grad_norm": 0.3544990534025258, + "learning_rate": 4.576775483166536e-05, + "loss": 0.3234, + "step": 1850 + }, + { + "epoch": 0.8100436681222707, + "grad_norm": 0.3459858077037921, + "learning_rate": 4.5732330000734775e-05, + "loss": 0.3319, + "step": 1855 + }, + { + "epoch": 0.8122270742358079, + "grad_norm": 0.3524884912271132, + "learning_rate": 4.569677135962864e-05, + "loss": 0.3278, + "step": 1860 + }, + { + "epoch": 0.8144104803493449, + "grad_norm": 0.42168404868370957, + "learning_rate": 4.5661079137848817e-05, + "loss": 0.314, + "step": 1865 + }, + { + "epoch": 0.8165938864628821, + "grad_norm": 0.4177554810132261, + "learning_rate": 4.562525356575928e-05, + "loss": 0.319, + "step": 1870 + }, + { + "epoch": 0.8187772925764192, + "grad_norm": 0.3406360499086555, + "learning_rate": 4.558929487458472e-05, + "loss": 0.3307, + "step": 1875 + }, + { + "epoch": 0.8209606986899564, + "grad_norm": 0.5953338048353011, + "learning_rate": 4.5553203296408966e-05, + "loss": 0.3363, + "step": 1880 + }, + { + "epoch": 0.8231441048034934, + "grad_norm": 0.3755495334670722, + "learning_rate": 4.551697906417353e-05, + "loss": 0.3228, + "step": 1885 + }, + { + "epoch": 0.8253275109170306, + "grad_norm": 0.5244192452428544, + "learning_rate": 4.5480622411676105e-05, + "loss": 0.3371, + "step": 1890 + }, + { + "epoch": 0.8275109170305677, + "grad_norm": 0.42718167146451547, + "learning_rate": 4.5444133573569056e-05, + "loss": 0.3394, + "step": 1895 + }, + { + "epoch": 0.8296943231441049, + "grad_norm": 0.5786026070553736, + "learning_rate": 4.540751278535788e-05, + "loss": 0.3413, + "step": 1900 + }, + { + "epoch": 0.8318777292576419, + "grad_norm": 0.3994255104166588, + "learning_rate": 4.5370760283399714e-05, + "loss": 0.3237, + "step": 1905 + }, + { + "epoch": 0.834061135371179, + "grad_norm": 0.465218260660747, + "learning_rate": 4.533387630490179e-05, + "loss": 0.3258, + "step": 1910 + }, + { + "epoch": 0.8362445414847162, + "grad_norm": 0.3503020390555209, + "learning_rate": 4.529686108791993e-05, + "loss": 0.3365, + "step": 1915 + }, + { + "epoch": 0.8384279475982532, + "grad_norm": 0.2937649633191906, + "learning_rate": 4.525971487135697e-05, + "loss": 0.3261, + "step": 1920 + }, + { + "epoch": 0.8406113537117904, + "grad_norm": 0.3640608075443947, + "learning_rate": 4.522243789496125e-05, + "loss": 0.3314, + "step": 1925 + }, + { + "epoch": 0.8427947598253275, + "grad_norm": 0.46642738830210834, + "learning_rate": 4.518503039932508e-05, + "loss": 0.3294, + "step": 1930 + }, + { + "epoch": 0.8449781659388647, + "grad_norm": 0.49375163248568255, + "learning_rate": 4.5147492625883133e-05, + "loss": 0.3346, + "step": 1935 + }, + { + "epoch": 0.8471615720524017, + "grad_norm": 0.43745545950003145, + "learning_rate": 4.510982481691093e-05, + "loss": 0.3356, + "step": 1940 + }, + { + "epoch": 0.8493449781659389, + "grad_norm": 0.3849745896242002, + "learning_rate": 4.5072027215523256e-05, + "loss": 0.328, + "step": 1945 + }, + { + "epoch": 0.851528384279476, + "grad_norm": 0.323723856108267, + "learning_rate": 4.503410006567262e-05, + "loss": 0.3137, + "step": 1950 + }, + { + "epoch": 0.8537117903930131, + "grad_norm": 0.3497043951204244, + "learning_rate": 4.499604361214764e-05, + "loss": 0.3367, + "step": 1955 + }, + { + "epoch": 0.8558951965065502, + "grad_norm": 0.3279206934648897, + "learning_rate": 4.495785810057151e-05, + "loss": 0.3305, + "step": 1960 + }, + { + "epoch": 0.8580786026200873, + "grad_norm": 0.35534435710755663, + "learning_rate": 4.4919543777400354e-05, + "loss": 0.3148, + "step": 1965 + }, + { + "epoch": 0.8602620087336245, + "grad_norm": 0.35737396766118557, + "learning_rate": 4.488110088992168e-05, + "loss": 0.3365, + "step": 1970 + }, + { + "epoch": 0.8624454148471615, + "grad_norm": 0.3811344665888169, + "learning_rate": 4.4842529686252774e-05, + "loss": 0.3412, + "step": 1975 + }, + { + "epoch": 0.8646288209606987, + "grad_norm": 0.3586035760646476, + "learning_rate": 4.4803830415339124e-05, + "loss": 0.3319, + "step": 1980 + }, + { + "epoch": 0.8668122270742358, + "grad_norm": 0.35951201381552045, + "learning_rate": 4.476500332695275e-05, + "loss": 0.3123, + "step": 1985 + }, + { + "epoch": 0.868995633187773, + "grad_norm": 0.4398889292154198, + "learning_rate": 4.472604867169063e-05, + "loss": 0.3258, + "step": 1990 + }, + { + "epoch": 0.87117903930131, + "grad_norm": 0.45684784821543106, + "learning_rate": 4.468696670097312e-05, + "loss": 0.3321, + "step": 1995 + }, + { + "epoch": 0.8733624454148472, + "grad_norm": 0.4624735089982277, + "learning_rate": 4.464775766704225e-05, + "loss": 0.339, + "step": 2000 + }, + { + "epoch": 0.8755458515283843, + "grad_norm": 0.39373793988338507, + "learning_rate": 4.4608421822960165e-05, + "loss": 0.3232, + "step": 2005 + }, + { + "epoch": 0.8777292576419214, + "grad_norm": 0.4793265990786189, + "learning_rate": 4.456895942260745e-05, + "loss": 0.3115, + "step": 2010 + }, + { + "epoch": 0.8799126637554585, + "grad_norm": 0.4596917524946721, + "learning_rate": 4.4529370720681516e-05, + "loss": 0.3327, + "step": 2015 + }, + { + "epoch": 0.8820960698689956, + "grad_norm": 0.41630714877910346, + "learning_rate": 4.4489655972694946e-05, + "loss": 0.3234, + "step": 2020 + }, + { + "epoch": 0.8842794759825328, + "grad_norm": 0.433659367959683, + "learning_rate": 4.444981543497384e-05, + "loss": 0.3268, + "step": 2025 + }, + { + "epoch": 0.8864628820960698, + "grad_norm": 0.4911763094493765, + "learning_rate": 4.440984936465618e-05, + "loss": 0.3283, + "step": 2030 + }, + { + "epoch": 0.888646288209607, + "grad_norm": 0.43509708304067723, + "learning_rate": 4.436975801969013e-05, + "loss": 0.3166, + "step": 2035 + }, + { + "epoch": 0.8908296943231441, + "grad_norm": 0.32181145619100665, + "learning_rate": 4.4329541658832454e-05, + "loss": 0.3275, + "step": 2040 + }, + { + "epoch": 0.8930131004366813, + "grad_norm": 0.5173193806980154, + "learning_rate": 4.428920054164673e-05, + "loss": 0.3367, + "step": 2045 + }, + { + "epoch": 0.8951965065502183, + "grad_norm": 0.31983525834019055, + "learning_rate": 4.424873492850177e-05, + "loss": 0.3332, + "step": 2050 + }, + { + "epoch": 0.8973799126637555, + "grad_norm": 0.3070335963036919, + "learning_rate": 4.42081450805699e-05, + "loss": 0.3416, + "step": 2055 + }, + { + "epoch": 0.8995633187772926, + "grad_norm": 0.33240317525320323, + "learning_rate": 4.4167431259825274e-05, + "loss": 0.3214, + "step": 2060 + }, + { + "epoch": 0.9017467248908297, + "grad_norm": 0.6261250657003081, + "learning_rate": 4.41265937290422e-05, + "loss": 0.3403, + "step": 2065 + }, + { + "epoch": 0.9039301310043668, + "grad_norm": 0.6469352618052645, + "learning_rate": 4.408563275179341e-05, + "loss": 0.3161, + "step": 2070 + }, + { + "epoch": 0.9061135371179039, + "grad_norm": 0.5128181692159853, + "learning_rate": 4.404454859244841e-05, + "loss": 0.3513, + "step": 2075 + }, + { + "epoch": 0.9082969432314411, + "grad_norm": 0.4054137512768172, + "learning_rate": 4.400334151617171e-05, + "loss": 0.3551, + "step": 2080 + }, + { + "epoch": 0.9104803493449781, + "grad_norm": 0.5160259798496531, + "learning_rate": 4.39620117889212e-05, + "loss": 0.3202, + "step": 2085 + }, + { + "epoch": 0.9126637554585153, + "grad_norm": 1.180298141482782, + "learning_rate": 4.3920559677446335e-05, + "loss": 0.3264, + "step": 2090 + }, + { + "epoch": 0.9148471615720524, + "grad_norm": 0.4807710172298594, + "learning_rate": 4.387898544928648e-05, + "loss": 0.3382, + "step": 2095 + }, + { + "epoch": 0.9170305676855895, + "grad_norm": 0.5325040390840762, + "learning_rate": 4.383728937276915e-05, + "loss": 0.3408, + "step": 2100 + }, + { + "epoch": 0.9192139737991266, + "grad_norm": 0.4423809835776037, + "learning_rate": 4.37954717170083e-05, + "loss": 0.3336, + "step": 2105 + }, + { + "epoch": 0.9213973799126638, + "grad_norm": 0.5498410601529987, + "learning_rate": 4.375353275190259e-05, + "loss": 0.3309, + "step": 2110 + }, + { + "epoch": 0.9235807860262009, + "grad_norm": 0.37076013082197373, + "learning_rate": 4.371147274813362e-05, + "loss": 0.3414, + "step": 2115 + }, + { + "epoch": 0.925764192139738, + "grad_norm": 0.403897556358072, + "learning_rate": 4.3669291977164185e-05, + "loss": 0.3308, + "step": 2120 + }, + { + "epoch": 0.9279475982532751, + "grad_norm": 0.4050123744090438, + "learning_rate": 4.362699071123655e-05, + "loss": 0.3383, + "step": 2125 + }, + { + "epoch": 0.9301310043668122, + "grad_norm": 0.4499616242086888, + "learning_rate": 4.358456922337068e-05, + "loss": 0.3247, + "step": 2130 + }, + { + "epoch": 0.9323144104803494, + "grad_norm": 0.41076931445914805, + "learning_rate": 4.354202778736245e-05, + "loss": 0.3325, + "step": 2135 + }, + { + "epoch": 0.9344978165938864, + "grad_norm": 0.6474266802788862, + "learning_rate": 4.3499366677781927e-05, + "loss": 0.3384, + "step": 2140 + }, + { + "epoch": 0.9366812227074236, + "grad_norm": 0.43958950888079396, + "learning_rate": 4.3456586169971555e-05, + "loss": 0.3155, + "step": 2145 + }, + { + "epoch": 0.9388646288209607, + "grad_norm": 0.430386825850428, + "learning_rate": 4.341368654004442e-05, + "loss": 0.3407, + "step": 2150 + }, + { + "epoch": 0.9410480349344978, + "grad_norm": 0.4400594923102947, + "learning_rate": 4.3370668064882406e-05, + "loss": 0.3227, + "step": 2155 + }, + { + "epoch": 0.9432314410480349, + "grad_norm": 0.49619465693365633, + "learning_rate": 4.332753102213447e-05, + "loss": 0.3488, + "step": 2160 + }, + { + "epoch": 0.9454148471615721, + "grad_norm": 0.2953653819485504, + "learning_rate": 4.328427569021481e-05, + "loss": 0.3237, + "step": 2165 + }, + { + "epoch": 0.9475982532751092, + "grad_norm": 0.3277990570496694, + "learning_rate": 4.3240902348301134e-05, + "loss": 0.3137, + "step": 2170 + }, + { + "epoch": 0.9497816593886463, + "grad_norm": 0.31994132171967327, + "learning_rate": 4.319741127633273e-05, + "loss": 0.3263, + "step": 2175 + }, + { + "epoch": 0.9519650655021834, + "grad_norm": 0.3159818487824089, + "learning_rate": 4.315380275500879e-05, + "loss": 0.3342, + "step": 2180 + }, + { + "epoch": 0.9541484716157205, + "grad_norm": 0.37710542197615304, + "learning_rate": 4.311007706578654e-05, + "loss": 0.3202, + "step": 2185 + }, + { + "epoch": 0.9563318777292577, + "grad_norm": 0.39548613497208596, + "learning_rate": 4.306623449087939e-05, + "loss": 0.3182, + "step": 2190 + }, + { + "epoch": 0.9585152838427947, + "grad_norm": 0.3737799555086228, + "learning_rate": 4.3022275313255215e-05, + "loss": 0.3202, + "step": 2195 + }, + { + "epoch": 0.9606986899563319, + "grad_norm": 0.3302761729103403, + "learning_rate": 4.2978199816634404e-05, + "loss": 0.3134, + "step": 2200 + }, + { + "epoch": 0.962882096069869, + "grad_norm": 0.3169568046490519, + "learning_rate": 4.293400828548812e-05, + "loss": 0.3281, + "step": 2205 + }, + { + "epoch": 0.9650655021834061, + "grad_norm": 0.3469828187440111, + "learning_rate": 4.2889701005036444e-05, + "loss": 0.3257, + "step": 2210 + }, + { + "epoch": 0.9672489082969432, + "grad_norm": 0.41804346589863545, + "learning_rate": 4.28452782612465e-05, + "loss": 0.3395, + "step": 2215 + }, + { + "epoch": 0.9694323144104804, + "grad_norm": 0.33364679303892797, + "learning_rate": 4.2800740340830636e-05, + "loss": 0.3127, + "step": 2220 + }, + { + "epoch": 0.9716157205240175, + "grad_norm": 0.3192103795670632, + "learning_rate": 4.275608753124459e-05, + "loss": 0.3139, + "step": 2225 + }, + { + "epoch": 0.9737991266375546, + "grad_norm": 0.43164296461188517, + "learning_rate": 4.271132012068559e-05, + "loss": 0.3144, + "step": 2230 + }, + { + "epoch": 0.9759825327510917, + "grad_norm": 0.36786016009963335, + "learning_rate": 4.266643839809054e-05, + "loss": 0.3217, + "step": 2235 + }, + { + "epoch": 0.9781659388646288, + "grad_norm": 0.40515456026599855, + "learning_rate": 4.262144265313412e-05, + "loss": 0.3263, + "step": 2240 + }, + { + "epoch": 0.980349344978166, + "grad_norm": 0.40316039408536825, + "learning_rate": 4.2576333176226944e-05, + "loss": 0.3174, + "step": 2245 + }, + { + "epoch": 0.982532751091703, + "grad_norm": 0.4036377506232831, + "learning_rate": 4.253111025851366e-05, + "loss": 0.3324, + "step": 2250 + }, + { + "epoch": 0.9847161572052402, + "grad_norm": 0.38600938504920373, + "learning_rate": 4.2485774191871084e-05, + "loss": 0.3148, + "step": 2255 + }, + { + "epoch": 0.9868995633187773, + "grad_norm": 0.3724866092235095, + "learning_rate": 4.244032526890633e-05, + "loss": 0.32, + "step": 2260 + }, + { + "epoch": 0.9890829694323144, + "grad_norm": 0.45130824685960164, + "learning_rate": 4.239476378295489e-05, + "loss": 0.3208, + "step": 2265 + }, + { + "epoch": 0.9912663755458515, + "grad_norm": 0.3614299023632802, + "learning_rate": 4.2349090028078766e-05, + "loss": 0.3108, + "step": 2270 + }, + { + "epoch": 0.9934497816593887, + "grad_norm": 0.3529974003854878, + "learning_rate": 4.230330429906456e-05, + "loss": 0.3114, + "step": 2275 + }, + { + "epoch": 0.9956331877729258, + "grad_norm": 0.3358112816928416, + "learning_rate": 4.225740689142158e-05, + "loss": 0.3168, + "step": 2280 + }, + { + "epoch": 0.9978165938864629, + "grad_norm": 0.3907341561097172, + "learning_rate": 4.221139810137992e-05, + "loss": 0.3174, + "step": 2285 + }, + { + "epoch": 1.0, + "grad_norm": 0.37377145244747273, + "learning_rate": 4.216527822588857e-05, + "loss": 0.3168, + "step": 2290 + }, + { + "epoch": 1.002183406113537, + "grad_norm": 0.40852537721961585, + "learning_rate": 4.2119047562613476e-05, + "loss": 0.2941, + "step": 2295 + }, + { + "epoch": 1.0043668122270741, + "grad_norm": 0.39878675388447493, + "learning_rate": 4.207270640993561e-05, + "loss": 0.2787, + "step": 2300 + }, + { + "epoch": 1.0065502183406114, + "grad_norm": 0.346225297182408, + "learning_rate": 4.202625506694911e-05, + "loss": 0.2686, + "step": 2305 + }, + { + "epoch": 1.0087336244541485, + "grad_norm": 0.34208833101355196, + "learning_rate": 4.197969383345925e-05, + "loss": 0.2809, + "step": 2310 + }, + { + "epoch": 1.0109170305676856, + "grad_norm": 0.4002016670275176, + "learning_rate": 4.19330230099806e-05, + "loss": 0.2979, + "step": 2315 + }, + { + "epoch": 1.0131004366812226, + "grad_norm": 0.4630743803988168, + "learning_rate": 4.188624289773501e-05, + "loss": 0.2967, + "step": 2320 + }, + { + "epoch": 1.01528384279476, + "grad_norm": 0.3568696275266243, + "learning_rate": 4.1839353798649717e-05, + "loss": 0.2758, + "step": 2325 + }, + { + "epoch": 1.017467248908297, + "grad_norm": 0.3119249400022314, + "learning_rate": 4.1792356015355375e-05, + "loss": 0.2766, + "step": 2330 + }, + { + "epoch": 1.019650655021834, + "grad_norm": 0.3696696936235827, + "learning_rate": 4.17452498511841e-05, + "loss": 0.286, + "step": 2335 + }, + { + "epoch": 1.0218340611353711, + "grad_norm": 0.3084313639844132, + "learning_rate": 4.169803561016754e-05, + "loss": 0.2856, + "step": 2340 + }, + { + "epoch": 1.0240174672489082, + "grad_norm": 0.3909495635210935, + "learning_rate": 4.1650713597034845e-05, + "loss": 0.287, + "step": 2345 + }, + { + "epoch": 1.0262008733624455, + "grad_norm": 0.44840025564529157, + "learning_rate": 4.16032841172108e-05, + "loss": 0.2927, + "step": 2350 + }, + { + "epoch": 1.0283842794759825, + "grad_norm": 0.41340997676020685, + "learning_rate": 4.155574747681375e-05, + "loss": 0.2858, + "step": 2355 + }, + { + "epoch": 1.0305676855895196, + "grad_norm": 0.3994213300123044, + "learning_rate": 4.1508103982653715e-05, + "loss": 0.2991, + "step": 2360 + }, + { + "epoch": 1.0327510917030567, + "grad_norm": 0.33838520158365676, + "learning_rate": 4.146035394223034e-05, + "loss": 0.2776, + "step": 2365 + }, + { + "epoch": 1.034934497816594, + "grad_norm": 0.3368322156288848, + "learning_rate": 4.1412497663730954e-05, + "loss": 0.2748, + "step": 2370 + }, + { + "epoch": 1.037117903930131, + "grad_norm": 0.3141352669129108, + "learning_rate": 4.1364535456028555e-05, + "loss": 0.2841, + "step": 2375 + }, + { + "epoch": 1.039301310043668, + "grad_norm": 0.2818967412459555, + "learning_rate": 4.1316467628679834e-05, + "loss": 0.2619, + "step": 2380 + }, + { + "epoch": 1.0414847161572052, + "grad_norm": 0.28565982519466604, + "learning_rate": 4.126829449192316e-05, + "loss": 0.2857, + "step": 2385 + }, + { + "epoch": 1.0436681222707425, + "grad_norm": 0.3210535970776655, + "learning_rate": 4.1220016356676615e-05, + "loss": 0.2787, + "step": 2390 + }, + { + "epoch": 1.0458515283842795, + "grad_norm": 0.3919954733516357, + "learning_rate": 4.117163353453593e-05, + "loss": 0.2824, + "step": 2395 + }, + { + "epoch": 1.0480349344978166, + "grad_norm": 0.2781986797396297, + "learning_rate": 4.112314633777251e-05, + "loss": 0.2831, + "step": 2400 + }, + { + "epoch": 1.0502183406113537, + "grad_norm": 0.3117143962813532, + "learning_rate": 4.107455507933144e-05, + "loss": 0.2741, + "step": 2405 + }, + { + "epoch": 1.0524017467248907, + "grad_norm": 0.29749854620448457, + "learning_rate": 4.102586007282938e-05, + "loss": 0.2784, + "step": 2410 + }, + { + "epoch": 1.054585152838428, + "grad_norm": 0.3011787067490516, + "learning_rate": 4.0977061632552656e-05, + "loss": 0.2822, + "step": 2415 + }, + { + "epoch": 1.056768558951965, + "grad_norm": 0.27733319238652004, + "learning_rate": 4.092816007345515e-05, + "loss": 0.2918, + "step": 2420 + }, + { + "epoch": 1.0589519650655022, + "grad_norm": 0.3178714706734469, + "learning_rate": 4.087915571115629e-05, + "loss": 0.2657, + "step": 2425 + }, + { + "epoch": 1.0611353711790392, + "grad_norm": 0.33147612300712626, + "learning_rate": 4.083004886193902e-05, + "loss": 0.2821, + "step": 2430 + }, + { + "epoch": 1.0633187772925765, + "grad_norm": 0.32015236720081114, + "learning_rate": 4.0780839842747754e-05, + "loss": 0.2876, + "step": 2435 + }, + { + "epoch": 1.0655021834061136, + "grad_norm": 0.2637393126437815, + "learning_rate": 4.073152897118632e-05, + "loss": 0.2902, + "step": 2440 + }, + { + "epoch": 1.0676855895196506, + "grad_norm": 0.2971292010190096, + "learning_rate": 4.0682116565515935e-05, + "loss": 0.294, + "step": 2445 + }, + { + "epoch": 1.0698689956331877, + "grad_norm": 0.32151195135466043, + "learning_rate": 4.063260294465311e-05, + "loss": 0.2754, + "step": 2450 + }, + { + "epoch": 1.072052401746725, + "grad_norm": 0.29344847997392354, + "learning_rate": 4.058298842816765e-05, + "loss": 0.2756, + "step": 2455 + }, + { + "epoch": 1.074235807860262, + "grad_norm": 0.38141464593456376, + "learning_rate": 4.0533273336280516e-05, + "loss": 0.2749, + "step": 2460 + }, + { + "epoch": 1.0764192139737991, + "grad_norm": 0.3519071752481233, + "learning_rate": 4.048345798986186e-05, + "loss": 0.2937, + "step": 2465 + }, + { + "epoch": 1.0786026200873362, + "grad_norm": 0.27974724944596935, + "learning_rate": 4.043354271042884e-05, + "loss": 0.2772, + "step": 2470 + }, + { + "epoch": 1.0807860262008733, + "grad_norm": 0.31883200998020006, + "learning_rate": 4.038352782014362e-05, + "loss": 0.2783, + "step": 2475 + }, + { + "epoch": 1.0829694323144106, + "grad_norm": 0.3107436150415206, + "learning_rate": 4.033341364181128e-05, + "loss": 0.2845, + "step": 2480 + }, + { + "epoch": 1.0851528384279476, + "grad_norm": 0.3631114446623769, + "learning_rate": 4.028320049887771e-05, + "loss": 0.2815, + "step": 2485 + }, + { + "epoch": 1.0873362445414847, + "grad_norm": 0.29440410304916975, + "learning_rate": 4.023288871542752e-05, + "loss": 0.2767, + "step": 2490 + }, + { + "epoch": 1.0895196506550218, + "grad_norm": 0.3013728586687223, + "learning_rate": 4.0182478616182015e-05, + "loss": 0.2802, + "step": 2495 + }, + { + "epoch": 1.091703056768559, + "grad_norm": 0.30015696374357026, + "learning_rate": 4.013197052649699e-05, + "loss": 0.2885, + "step": 2500 + }, + { + "epoch": 1.0938864628820961, + "grad_norm": 0.334054468972433, + "learning_rate": 4.0081364772360725e-05, + "loss": 0.2849, + "step": 2505 + }, + { + "epoch": 1.0960698689956332, + "grad_norm": 0.29580484838617715, + "learning_rate": 4.003066168039183e-05, + "loss": 0.2748, + "step": 2510 + }, + { + "epoch": 1.0982532751091703, + "grad_norm": 0.3454428161947495, + "learning_rate": 3.9979861577837155e-05, + "loss": 0.2783, + "step": 2515 + }, + { + "epoch": 1.1004366812227073, + "grad_norm": 0.3354653145039301, + "learning_rate": 3.9928964792569655e-05, + "loss": 0.2811, + "step": 2520 + }, + { + "epoch": 1.1026200873362446, + "grad_norm": 0.3257148621799395, + "learning_rate": 3.987797165308633e-05, + "loss": 0.2805, + "step": 2525 + }, + { + "epoch": 1.1048034934497817, + "grad_norm": 0.3093405188202137, + "learning_rate": 3.982688248850604e-05, + "loss": 0.2929, + "step": 2530 + }, + { + "epoch": 1.1069868995633187, + "grad_norm": 0.42822993107757057, + "learning_rate": 3.977569762856741e-05, + "loss": 0.2913, + "step": 2535 + }, + { + "epoch": 1.1091703056768558, + "grad_norm": 0.40323824571841865, + "learning_rate": 3.9724417403626705e-05, + "loss": 0.2834, + "step": 2540 + }, + { + "epoch": 1.111353711790393, + "grad_norm": 0.30813240614187476, + "learning_rate": 3.9673042144655697e-05, + "loss": 0.2927, + "step": 2545 + }, + { + "epoch": 1.1135371179039302, + "grad_norm": 0.2807299601361123, + "learning_rate": 3.9621572183239515e-05, + "loss": 0.2778, + "step": 2550 + }, + { + "epoch": 1.1157205240174672, + "grad_norm": 0.34539465437923245, + "learning_rate": 3.957000785157452e-05, + "loss": 0.3014, + "step": 2555 + }, + { + "epoch": 1.1179039301310043, + "grad_norm": 0.36001154980798994, + "learning_rate": 3.951834948246616e-05, + "loss": 0.268, + "step": 2560 + }, + { + "epoch": 1.1200873362445414, + "grad_norm": 0.3659504200165266, + "learning_rate": 3.946659740932681e-05, + "loss": 0.2999, + "step": 2565 + }, + { + "epoch": 1.1222707423580787, + "grad_norm": 0.310758158636425, + "learning_rate": 3.9414751966173634e-05, + "loss": 0.2788, + "step": 2570 + }, + { + "epoch": 1.1244541484716157, + "grad_norm": 0.2658948188578196, + "learning_rate": 3.93628134876264e-05, + "loss": 0.2764, + "step": 2575 + }, + { + "epoch": 1.1266375545851528, + "grad_norm": 0.3038259778170334, + "learning_rate": 3.9310782308905395e-05, + "loss": 0.2809, + "step": 2580 + }, + { + "epoch": 1.1288209606986899, + "grad_norm": 0.3098003775343629, + "learning_rate": 3.925865876582915e-05, + "loss": 0.2668, + "step": 2585 + }, + { + "epoch": 1.1310043668122272, + "grad_norm": 0.29478313929574995, + "learning_rate": 3.920644319481237e-05, + "loss": 0.2762, + "step": 2590 + }, + { + "epoch": 1.1331877729257642, + "grad_norm": 0.2715517904632468, + "learning_rate": 3.915413593286371e-05, + "loss": 0.2828, + "step": 2595 + }, + { + "epoch": 1.1353711790393013, + "grad_norm": 0.2546601742913032, + "learning_rate": 3.910173731758362e-05, + "loss": 0.2752, + "step": 2600 + }, + { + "epoch": 1.1375545851528384, + "grad_norm": 0.3020454024382658, + "learning_rate": 3.904924768716216e-05, + "loss": 0.2764, + "step": 2605 + }, + { + "epoch": 1.1397379912663754, + "grad_norm": 0.3404890082601122, + "learning_rate": 3.89966673803768e-05, + "loss": 0.2746, + "step": 2610 + }, + { + "epoch": 1.1419213973799127, + "grad_norm": 0.2918728991626841, + "learning_rate": 3.894399673659031e-05, + "loss": 0.2911, + "step": 2615 + }, + { + "epoch": 1.1441048034934498, + "grad_norm": 0.32777823728250605, + "learning_rate": 3.889123609574843e-05, + "loss": 0.2687, + "step": 2620 + }, + { + "epoch": 1.1462882096069869, + "grad_norm": 0.3417220224406648, + "learning_rate": 3.883838579837781e-05, + "loss": 0.29, + "step": 2625 + }, + { + "epoch": 1.1484716157205241, + "grad_norm": 0.3206470787242044, + "learning_rate": 3.8785446185583755e-05, + "loss": 0.2852, + "step": 2630 + }, + { + "epoch": 1.1506550218340612, + "grad_norm": 0.2896917429374818, + "learning_rate": 3.873241759904801e-05, + "loss": 0.2844, + "step": 2635 + }, + { + "epoch": 1.1528384279475983, + "grad_norm": 0.319012549688518, + "learning_rate": 3.8679300381026585e-05, + "loss": 0.2926, + "step": 2640 + }, + { + "epoch": 1.1550218340611353, + "grad_norm": 0.27491269743019364, + "learning_rate": 3.862609487434752e-05, + "loss": 0.2701, + "step": 2645 + }, + { + "epoch": 1.1572052401746724, + "grad_norm": 0.30736358656559387, + "learning_rate": 3.857280142240871e-05, + "loss": 0.2813, + "step": 2650 + }, + { + "epoch": 1.1593886462882097, + "grad_norm": 0.27847789277867213, + "learning_rate": 3.851942036917565e-05, + "loss": 0.2989, + "step": 2655 + }, + { + "epoch": 1.1615720524017468, + "grad_norm": 0.3178443314858301, + "learning_rate": 3.846595205917922e-05, + "loss": 0.2831, + "step": 2660 + }, + { + "epoch": 1.1637554585152838, + "grad_norm": 0.27407489870985063, + "learning_rate": 3.8412396837513484e-05, + "loss": 0.2838, + "step": 2665 + }, + { + "epoch": 1.165938864628821, + "grad_norm": 0.31079163261067777, + "learning_rate": 3.835875504983344e-05, + "loss": 0.2896, + "step": 2670 + }, + { + "epoch": 1.1681222707423582, + "grad_norm": 0.330388527773976, + "learning_rate": 3.830502704235282e-05, + "loss": 0.281, + "step": 2675 + }, + { + "epoch": 1.1703056768558953, + "grad_norm": 0.28816900411883056, + "learning_rate": 3.825121316184181e-05, + "loss": 0.2883, + "step": 2680 + }, + { + "epoch": 1.1724890829694323, + "grad_norm": 0.3328938276296137, + "learning_rate": 3.8197313755624855e-05, + "loss": 0.266, + "step": 2685 + }, + { + "epoch": 1.1746724890829694, + "grad_norm": 0.3824721867696784, + "learning_rate": 3.8143329171578366e-05, + "loss": 0.2792, + "step": 2690 + }, + { + "epoch": 1.1768558951965065, + "grad_norm": 0.33578112498653234, + "learning_rate": 3.808925975812855e-05, + "loss": 0.2994, + "step": 2695 + }, + { + "epoch": 1.1790393013100438, + "grad_norm": 0.35312074931486553, + "learning_rate": 3.803510586424908e-05, + "loss": 0.2909, + "step": 2700 + }, + { + "epoch": 1.1812227074235808, + "grad_norm": 0.3016960624575352, + "learning_rate": 3.798086783945892e-05, + "loss": 0.2829, + "step": 2705 + }, + { + "epoch": 1.1834061135371179, + "grad_norm": 0.355359254971825, + "learning_rate": 3.792654603381998e-05, + "loss": 0.2843, + "step": 2710 + }, + { + "epoch": 1.185589519650655, + "grad_norm": 0.3700674234618425, + "learning_rate": 3.7872140797934956e-05, + "loss": 0.2719, + "step": 2715 + }, + { + "epoch": 1.1877729257641922, + "grad_norm": 0.44056476547643436, + "learning_rate": 3.781765248294499e-05, + "loss": 0.2759, + "step": 2720 + }, + { + "epoch": 1.1899563318777293, + "grad_norm": 0.29079420170197845, + "learning_rate": 3.7763081440527435e-05, + "loss": 0.2857, + "step": 2725 + }, + { + "epoch": 1.1921397379912664, + "grad_norm": 0.312707282955329, + "learning_rate": 3.7708428022893596e-05, + "loss": 0.2863, + "step": 2730 + }, + { + "epoch": 1.1943231441048034, + "grad_norm": 0.2684008615845062, + "learning_rate": 3.7653692582786416e-05, + "loss": 0.272, + "step": 2735 + }, + { + "epoch": 1.1965065502183405, + "grad_norm": 0.35782293675245397, + "learning_rate": 3.7598875473478246e-05, + "loss": 0.2715, + "step": 2740 + }, + { + "epoch": 1.1986899563318778, + "grad_norm": 0.30301767808558216, + "learning_rate": 3.7543977048768534e-05, + "loss": 0.271, + "step": 2745 + }, + { + "epoch": 1.2008733624454149, + "grad_norm": 0.3151321773465957, + "learning_rate": 3.748899766298157e-05, + "loss": 0.2845, + "step": 2750 + }, + { + "epoch": 1.203056768558952, + "grad_norm": 0.2925859858510235, + "learning_rate": 3.743393767096416e-05, + "loss": 0.2755, + "step": 2755 + }, + { + "epoch": 1.205240174672489, + "grad_norm": 0.2859805443710802, + "learning_rate": 3.737879742808337e-05, + "loss": 0.2793, + "step": 2760 + }, + { + "epoch": 1.2074235807860263, + "grad_norm": 0.25461387164265176, + "learning_rate": 3.732357729022421e-05, + "loss": 0.2877, + "step": 2765 + }, + { + "epoch": 1.2096069868995634, + "grad_norm": 0.24813193774999823, + "learning_rate": 3.7268277613787356e-05, + "loss": 0.2869, + "step": 2770 + }, + { + "epoch": 1.2117903930131004, + "grad_norm": 0.3276586080798546, + "learning_rate": 3.721289875568683e-05, + "loss": 0.2853, + "step": 2775 + }, + { + "epoch": 1.2139737991266375, + "grad_norm": 0.3351580255362412, + "learning_rate": 3.715744107334773e-05, + "loss": 0.2699, + "step": 2780 + }, + { + "epoch": 1.2161572052401746, + "grad_norm": 0.30882004844208494, + "learning_rate": 3.7101904924703865e-05, + "loss": 0.2823, + "step": 2785 + }, + { + "epoch": 1.2183406113537119, + "grad_norm": 0.37760317246667774, + "learning_rate": 3.704629066819549e-05, + "loss": 0.2771, + "step": 2790 + }, + { + "epoch": 1.220524017467249, + "grad_norm": 0.32894843491571224, + "learning_rate": 3.6990598662767014e-05, + "loss": 0.2699, + "step": 2795 + }, + { + "epoch": 1.222707423580786, + "grad_norm": 0.3476225470830814, + "learning_rate": 3.693482926786461e-05, + "loss": 0.2726, + "step": 2800 + }, + { + "epoch": 1.224890829694323, + "grad_norm": 0.3118303300626587, + "learning_rate": 3.687898284343396e-05, + "loss": 0.2891, + "step": 2805 + }, + { + "epoch": 1.2270742358078603, + "grad_norm": 0.3281515137713933, + "learning_rate": 3.68230597499179e-05, + "loss": 0.2805, + "step": 2810 + }, + { + "epoch": 1.2292576419213974, + "grad_norm": 0.281506457906696, + "learning_rate": 3.676706034825409e-05, + "loss": 0.3019, + "step": 2815 + }, + { + "epoch": 1.2314410480349345, + "grad_norm": 0.2918564930941491, + "learning_rate": 3.6710984999872735e-05, + "loss": 0.2748, + "step": 2820 + }, + { + "epoch": 1.2336244541484715, + "grad_norm": 0.301158978005547, + "learning_rate": 3.665483406669418e-05, + "loss": 0.2904, + "step": 2825 + }, + { + "epoch": 1.2358078602620086, + "grad_norm": 0.29117932725379103, + "learning_rate": 3.6598607911126614e-05, + "loss": 0.2856, + "step": 2830 + }, + { + "epoch": 1.237991266375546, + "grad_norm": 0.2736978249597251, + "learning_rate": 3.654230689606372e-05, + "loss": 0.2659, + "step": 2835 + }, + { + "epoch": 1.240174672489083, + "grad_norm": 0.2595195793059447, + "learning_rate": 3.648593138488235e-05, + "loss": 0.278, + "step": 2840 + }, + { + "epoch": 1.24235807860262, + "grad_norm": 0.31138011895796025, + "learning_rate": 3.6429481741440176e-05, + "loss": 0.2804, + "step": 2845 + }, + { + "epoch": 1.244541484716157, + "grad_norm": 0.28491002649715325, + "learning_rate": 3.63729583300733e-05, + "loss": 0.278, + "step": 2850 + }, + { + "epoch": 1.2467248908296944, + "grad_norm": 0.28329885485362183, + "learning_rate": 3.6316361515593966e-05, + "loss": 0.2874, + "step": 2855 + }, + { + "epoch": 1.2489082969432315, + "grad_norm": 0.36368480663196673, + "learning_rate": 3.625969166328815e-05, + "loss": 0.276, + "step": 2860 + }, + { + "epoch": 1.2510917030567685, + "grad_norm": 0.2700235560170439, + "learning_rate": 3.620294913891326e-05, + "loss": 0.2884, + "step": 2865 + }, + { + "epoch": 1.2532751091703056, + "grad_norm": 0.25418847496227326, + "learning_rate": 3.614613430869571e-05, + "loss": 0.2859, + "step": 2870 + }, + { + "epoch": 1.2554585152838427, + "grad_norm": 0.28230477688636335, + "learning_rate": 3.608924753932862e-05, + "loss": 0.2884, + "step": 2875 + }, + { + "epoch": 1.25764192139738, + "grad_norm": 0.31064202314287603, + "learning_rate": 3.6032289197969374e-05, + "loss": 0.2858, + "step": 2880 + }, + { + "epoch": 1.259825327510917, + "grad_norm": 0.272199049103352, + "learning_rate": 3.5975259652237333e-05, + "loss": 0.287, + "step": 2885 + }, + { + "epoch": 1.262008733624454, + "grad_norm": 0.23256991109619604, + "learning_rate": 3.5918159270211424e-05, + "loss": 0.2734, + "step": 2890 + }, + { + "epoch": 1.2641921397379914, + "grad_norm": 0.2926902120837456, + "learning_rate": 3.586098842042773e-05, + "loss": 0.2774, + "step": 2895 + }, + { + "epoch": 1.2663755458515285, + "grad_norm": 0.36624779554626996, + "learning_rate": 3.580374747187719e-05, + "loss": 0.2775, + "step": 2900 + }, + { + "epoch": 1.2685589519650655, + "grad_norm": 0.29331268838279223, + "learning_rate": 3.574643679400313e-05, + "loss": 0.2747, + "step": 2905 + }, + { + "epoch": 1.2707423580786026, + "grad_norm": 0.27621174524527253, + "learning_rate": 3.568905675669895e-05, + "loss": 0.2882, + "step": 2910 + }, + { + "epoch": 1.2729257641921397, + "grad_norm": 0.2771703632616136, + "learning_rate": 3.5631607730305694e-05, + "loss": 0.2664, + "step": 2915 + }, + { + "epoch": 1.2751091703056767, + "grad_norm": 0.33588301272251375, + "learning_rate": 3.5574090085609696e-05, + "loss": 0.2798, + "step": 2920 + }, + { + "epoch": 1.277292576419214, + "grad_norm": 0.35009121663132425, + "learning_rate": 3.551650419384013e-05, + "loss": 0.2833, + "step": 2925 + }, + { + "epoch": 1.279475982532751, + "grad_norm": 0.2687807965233062, + "learning_rate": 3.545885042666669e-05, + "loss": 0.2738, + "step": 2930 + }, + { + "epoch": 1.2816593886462881, + "grad_norm": 0.36331931036035, + "learning_rate": 3.5401129156197105e-05, + "loss": 0.2833, + "step": 2935 + }, + { + "epoch": 1.2838427947598254, + "grad_norm": 0.3828328306666377, + "learning_rate": 3.534334075497482e-05, + "loss": 0.2821, + "step": 2940 + }, + { + "epoch": 1.2860262008733625, + "grad_norm": 0.30473362904993256, + "learning_rate": 3.5285485595976534e-05, + "loss": 0.2821, + "step": 2945 + }, + { + "epoch": 1.2882096069868996, + "grad_norm": 0.3499861198156561, + "learning_rate": 3.5227564052609816e-05, + "loss": 0.2724, + "step": 2950 + }, + { + "epoch": 1.2903930131004366, + "grad_norm": 0.2938833486098817, + "learning_rate": 3.5169576498710696e-05, + "loss": 0.278, + "step": 2955 + }, + { + "epoch": 1.2925764192139737, + "grad_norm": 0.28084189095150464, + "learning_rate": 3.511152330854125e-05, + "loss": 0.2679, + "step": 2960 + }, + { + "epoch": 1.294759825327511, + "grad_norm": 0.3199421686229126, + "learning_rate": 3.505340485678717e-05, + "loss": 0.2819, + "step": 2965 + }, + { + "epoch": 1.296943231441048, + "grad_norm": 0.3149994117491474, + "learning_rate": 3.499522151855537e-05, + "loss": 0.2975, + "step": 2970 + }, + { + "epoch": 1.2991266375545851, + "grad_norm": 0.2899693019311649, + "learning_rate": 3.493697366937154e-05, + "loss": 0.2706, + "step": 2975 + }, + { + "epoch": 1.3013100436681222, + "grad_norm": 0.33913499909626604, + "learning_rate": 3.487866168517775e-05, + "loss": 0.294, + "step": 2980 + }, + { + "epoch": 1.3034934497816595, + "grad_norm": 0.3769147657930795, + "learning_rate": 3.4820285942330003e-05, + "loss": 0.2818, + "step": 2985 + }, + { + "epoch": 1.3056768558951966, + "grad_norm": 0.32801615796861155, + "learning_rate": 3.476184681759581e-05, + "loss": 0.28, + "step": 2990 + }, + { + "epoch": 1.3078602620087336, + "grad_norm": 0.27357707367933753, + "learning_rate": 3.4703344688151776e-05, + "loss": 0.2783, + "step": 2995 + }, + { + "epoch": 1.3100436681222707, + "grad_norm": 0.37622537438391995, + "learning_rate": 3.46447799315811e-05, + "loss": 0.2819, + "step": 3000 + }, + { + "epoch": 1.3122270742358078, + "grad_norm": 0.3550543812495112, + "learning_rate": 3.458615292587125e-05, + "loss": 0.2792, + "step": 3005 + }, + { + "epoch": 1.314410480349345, + "grad_norm": 0.2816496834529316, + "learning_rate": 3.452746404941143e-05, + "loss": 0.2859, + "step": 3010 + }, + { + "epoch": 1.3165938864628821, + "grad_norm": 0.3066581463920424, + "learning_rate": 3.446871368099017e-05, + "loss": 0.2746, + "step": 3015 + }, + { + "epoch": 1.3187772925764192, + "grad_norm": 0.2708116400301654, + "learning_rate": 3.440990219979289e-05, + "loss": 0.2863, + "step": 3020 + }, + { + "epoch": 1.3209606986899565, + "grad_norm": 0.2497661453451677, + "learning_rate": 3.435102998539942e-05, + "loss": 0.2702, + "step": 3025 + }, + { + "epoch": 1.3231441048034935, + "grad_norm": 0.2651767333352664, + "learning_rate": 3.4292097417781595e-05, + "loss": 0.2695, + "step": 3030 + }, + { + "epoch": 1.3253275109170306, + "grad_norm": 0.2695245871370125, + "learning_rate": 3.4233104877300775e-05, + "loss": 0.267, + "step": 3035 + }, + { + "epoch": 1.3275109170305677, + "grad_norm": 0.27077504903552757, + "learning_rate": 3.41740527447054e-05, + "loss": 0.2662, + "step": 3040 + }, + { + "epoch": 1.3296943231441047, + "grad_norm": 0.2519617614699693, + "learning_rate": 3.411494140112852e-05, + "loss": 0.2716, + "step": 3045 + }, + { + "epoch": 1.3318777292576418, + "grad_norm": 0.2809728902320245, + "learning_rate": 3.4055771228085347e-05, + "loss": 0.2879, + "step": 3050 + }, + { + "epoch": 1.334061135371179, + "grad_norm": 0.26727497601257166, + "learning_rate": 3.399654260747078e-05, + "loss": 0.2725, + "step": 3055 + }, + { + "epoch": 1.3362445414847162, + "grad_norm": 0.34198746633106286, + "learning_rate": 3.3937255921556974e-05, + "loss": 0.2877, + "step": 3060 + }, + { + "epoch": 1.3384279475982532, + "grad_norm": 0.2984040980470414, + "learning_rate": 3.3877911552990826e-05, + "loss": 0.2793, + "step": 3065 + }, + { + "epoch": 1.3406113537117905, + "grad_norm": 0.2865962833130215, + "learning_rate": 3.3818509884791514e-05, + "loss": 0.273, + "step": 3070 + }, + { + "epoch": 1.3427947598253276, + "grad_norm": 0.2988181873287166, + "learning_rate": 3.375905130034808e-05, + "loss": 0.275, + "step": 3075 + }, + { + "epoch": 1.3449781659388647, + "grad_norm": 0.29108551078775924, + "learning_rate": 3.369953618341688e-05, + "loss": 0.2874, + "step": 3080 + }, + { + "epoch": 1.3471615720524017, + "grad_norm": 0.29395564574021693, + "learning_rate": 3.3639964918119135e-05, + "loss": 0.2756, + "step": 3085 + }, + { + "epoch": 1.3493449781659388, + "grad_norm": 0.2834037344232317, + "learning_rate": 3.3580337888938515e-05, + "loss": 0.2795, + "step": 3090 + }, + { + "epoch": 1.3515283842794759, + "grad_norm": 0.2963729779205327, + "learning_rate": 3.3520655480718515e-05, + "loss": 0.2821, + "step": 3095 + }, + { + "epoch": 1.3537117903930131, + "grad_norm": 0.2719250597184558, + "learning_rate": 3.346091807866013e-05, + "loss": 0.2799, + "step": 3100 + }, + { + "epoch": 1.3558951965065502, + "grad_norm": 0.345480183079055, + "learning_rate": 3.3401126068319244e-05, + "loss": 0.2753, + "step": 3105 + }, + { + "epoch": 1.3580786026200873, + "grad_norm": 0.2731322103388952, + "learning_rate": 3.3341279835604224e-05, + "loss": 0.2806, + "step": 3110 + }, + { + "epoch": 1.3602620087336246, + "grad_norm": 0.3377724432178437, + "learning_rate": 3.328137976677339e-05, + "loss": 0.2952, + "step": 3115 + }, + { + "epoch": 1.3624454148471616, + "grad_norm": 0.3490702531757625, + "learning_rate": 3.322142624843253e-05, + "loss": 0.2883, + "step": 3120 + }, + { + "epoch": 1.3646288209606987, + "grad_norm": 0.23733767943987039, + "learning_rate": 3.3161419667532384e-05, + "loss": 0.2811, + "step": 3125 + }, + { + "epoch": 1.3668122270742358, + "grad_norm": 0.29640032968995195, + "learning_rate": 3.3101360411366205e-05, + "loss": 0.2714, + "step": 3130 + }, + { + "epoch": 1.3689956331877728, + "grad_norm": 0.38335436641971393, + "learning_rate": 3.304124886756719e-05, + "loss": 0.2773, + "step": 3135 + }, + { + "epoch": 1.37117903930131, + "grad_norm": 0.2614713678332863, + "learning_rate": 3.2981085424106036e-05, + "loss": 0.2716, + "step": 3140 + }, + { + "epoch": 1.3733624454148472, + "grad_norm": 0.271556464815005, + "learning_rate": 3.2920870469288375e-05, + "loss": 0.2929, + "step": 3145 + }, + { + "epoch": 1.3755458515283843, + "grad_norm": 0.2984513438432199, + "learning_rate": 3.2860604391752344e-05, + "loss": 0.2767, + "step": 3150 + }, + { + "epoch": 1.3777292576419213, + "grad_norm": 0.35226423449647926, + "learning_rate": 3.2800287580466e-05, + "loss": 0.2803, + "step": 3155 + }, + { + "epoch": 1.3799126637554586, + "grad_norm": 0.2762721435202866, + "learning_rate": 3.273992042472487e-05, + "loss": 0.2635, + "step": 3160 + }, + { + "epoch": 1.3820960698689957, + "grad_norm": 0.2849399596588635, + "learning_rate": 3.2679503314149395e-05, + "loss": 0.2828, + "step": 3165 + }, + { + "epoch": 1.3842794759825328, + "grad_norm": 0.3240823060545034, + "learning_rate": 3.261903663868243e-05, + "loss": 0.2772, + "step": 3170 + }, + { + "epoch": 1.3864628820960698, + "grad_norm": 0.29529588632218107, + "learning_rate": 3.2558520788586764e-05, + "loss": 0.2709, + "step": 3175 + }, + { + "epoch": 1.388646288209607, + "grad_norm": 0.2880598231126244, + "learning_rate": 3.249795615444253e-05, + "loss": 0.2649, + "step": 3180 + }, + { + "epoch": 1.3908296943231442, + "grad_norm": 0.34085771304801965, + "learning_rate": 3.243734312714475e-05, + "loss": 0.2767, + "step": 3185 + }, + { + "epoch": 1.3930131004366813, + "grad_norm": 0.24864813820746293, + "learning_rate": 3.2376682097900754e-05, + "loss": 0.2793, + "step": 3190 + }, + { + "epoch": 1.3951965065502183, + "grad_norm": 0.3059620331889871, + "learning_rate": 3.2315973458227714e-05, + "loss": 0.2875, + "step": 3195 + }, + { + "epoch": 1.3973799126637554, + "grad_norm": 0.24526372956599277, + "learning_rate": 3.225521759995007e-05, + "loss": 0.2821, + "step": 3200 + }, + { + "epoch": 1.3995633187772927, + "grad_norm": 0.2972629377196201, + "learning_rate": 3.219441491519703e-05, + "loss": 0.2742, + "step": 3205 + }, + { + "epoch": 1.4017467248908297, + "grad_norm": 0.30368222569014375, + "learning_rate": 3.2133565796400046e-05, + "loss": 0.2794, + "step": 3210 + }, + { + "epoch": 1.4039301310043668, + "grad_norm": 0.27087580419910157, + "learning_rate": 3.2072670636290195e-05, + "loss": 0.2686, + "step": 3215 + }, + { + "epoch": 1.4061135371179039, + "grad_norm": 0.3097415372011331, + "learning_rate": 3.20117298278958e-05, + "loss": 0.2744, + "step": 3220 + }, + { + "epoch": 1.408296943231441, + "grad_norm": 0.265200535542659, + "learning_rate": 3.195074376453975e-05, + "loss": 0.2904, + "step": 3225 + }, + { + "epoch": 1.4104803493449782, + "grad_norm": 0.26447053395871756, + "learning_rate": 3.188971283983703e-05, + "loss": 0.2633, + "step": 3230 + }, + { + "epoch": 1.4126637554585153, + "grad_norm": 0.25630242114154234, + "learning_rate": 3.1828637447692187e-05, + "loss": 0.2755, + "step": 3235 + }, + { + "epoch": 1.4148471615720524, + "grad_norm": 0.26664274806120797, + "learning_rate": 3.176751798229673e-05, + "loss": 0.2802, + "step": 3240 + }, + { + "epoch": 1.4170305676855894, + "grad_norm": 0.3278108102940346, + "learning_rate": 3.1706354838126687e-05, + "loss": 0.2585, + "step": 3245 + }, + { + "epoch": 1.4192139737991267, + "grad_norm": 0.24680968950816637, + "learning_rate": 3.1645148409939915e-05, + "loss": 0.2758, + "step": 3250 + }, + { + "epoch": 1.4213973799126638, + "grad_norm": 0.2843693481761553, + "learning_rate": 3.158389909277371e-05, + "loss": 0.2857, + "step": 3255 + }, + { + "epoch": 1.4235807860262009, + "grad_norm": 0.27926774884591915, + "learning_rate": 3.152260728194213e-05, + "loss": 0.2827, + "step": 3260 + }, + { + "epoch": 1.425764192139738, + "grad_norm": 0.33114453429263857, + "learning_rate": 3.1461273373033516e-05, + "loss": 0.2791, + "step": 3265 + }, + { + "epoch": 1.427947598253275, + "grad_norm": 0.2523111291701471, + "learning_rate": 3.1399897761907914e-05, + "loss": 0.2785, + "step": 3270 + }, + { + "epoch": 1.4301310043668123, + "grad_norm": 0.2669982957777147, + "learning_rate": 3.133848084469454e-05, + "loss": 0.2723, + "step": 3275 + }, + { + "epoch": 1.4323144104803494, + "grad_norm": 0.2551505338781039, + "learning_rate": 3.1277023017789164e-05, + "loss": 0.2701, + "step": 3280 + }, + { + "epoch": 1.4344978165938864, + "grad_norm": 0.2729328232441411, + "learning_rate": 3.121552467785164e-05, + "loss": 0.2756, + "step": 3285 + }, + { + "epoch": 1.4366812227074237, + "grad_norm": 0.31413945020452155, + "learning_rate": 3.1153986221803284e-05, + "loss": 0.2634, + "step": 3290 + }, + { + "epoch": 1.4388646288209608, + "grad_norm": 0.27459622465012024, + "learning_rate": 3.109240804682431e-05, + "loss": 0.2749, + "step": 3295 + }, + { + "epoch": 1.4410480349344978, + "grad_norm": 0.3428676752152493, + "learning_rate": 3.103079055035131e-05, + "loss": 0.2737, + "step": 3300 + }, + { + "epoch": 1.443231441048035, + "grad_norm": 0.28826863252341717, + "learning_rate": 3.096913413007467e-05, + "loss": 0.2716, + "step": 3305 + }, + { + "epoch": 1.445414847161572, + "grad_norm": 0.3003661345160089, + "learning_rate": 3.090743918393596e-05, + "loss": 0.2711, + "step": 3310 + }, + { + "epoch": 1.447598253275109, + "grad_norm": 0.2726421109267964, + "learning_rate": 3.0845706110125435e-05, + "loss": 0.2755, + "step": 3315 + }, + { + "epoch": 1.4497816593886463, + "grad_norm": 0.2633565745117208, + "learning_rate": 3.078393530707942e-05, + "loss": 0.2708, + "step": 3320 + }, + { + "epoch": 1.4519650655021834, + "grad_norm": 0.3590173494625973, + "learning_rate": 3.072212717347776e-05, + "loss": 0.2802, + "step": 3325 + }, + { + "epoch": 1.4541484716157205, + "grad_norm": 0.2841165489479175, + "learning_rate": 3.0660282108241235e-05, + "loss": 0.2793, + "step": 3330 + }, + { + "epoch": 1.4563318777292578, + "grad_norm": 0.27787135238996175, + "learning_rate": 3.059840051052898e-05, + "loss": 0.2915, + "step": 3335 + }, + { + "epoch": 1.4585152838427948, + "grad_norm": 0.29883394182548156, + "learning_rate": 3.053648277973592e-05, + "loss": 0.2667, + "step": 3340 + }, + { + "epoch": 1.460698689956332, + "grad_norm": 0.2897989278866621, + "learning_rate": 3.04745293154902e-05, + "loss": 0.2863, + "step": 3345 + }, + { + "epoch": 1.462882096069869, + "grad_norm": 0.2413426041774491, + "learning_rate": 3.0412540517650582e-05, + "loss": 0.2686, + "step": 3350 + }, + { + "epoch": 1.465065502183406, + "grad_norm": 0.3092426474858127, + "learning_rate": 3.03505167863039e-05, + "loss": 0.2861, + "step": 3355 + }, + { + "epoch": 1.467248908296943, + "grad_norm": 0.37271590450228287, + "learning_rate": 3.0288458521762414e-05, + "loss": 0.2804, + "step": 3360 + }, + { + "epoch": 1.4694323144104804, + "grad_norm": 0.32074172110600374, + "learning_rate": 3.02263661245613e-05, + "loss": 0.2673, + "step": 3365 + }, + { + "epoch": 1.4716157205240175, + "grad_norm": 0.26878027230807305, + "learning_rate": 3.016423999545603e-05, + "loss": 0.2805, + "step": 3370 + }, + { + "epoch": 1.4737991266375545, + "grad_norm": 0.2611222636406977, + "learning_rate": 3.0102080535419775e-05, + "loss": 0.2852, + "step": 3375 + }, + { + "epoch": 1.4759825327510918, + "grad_norm": 0.32711140245362424, + "learning_rate": 3.0039888145640845e-05, + "loss": 0.2768, + "step": 3380 + }, + { + "epoch": 1.4781659388646289, + "grad_norm": 0.2933942315221079, + "learning_rate": 2.9977663227520065e-05, + "loss": 0.2733, + "step": 3385 + }, + { + "epoch": 1.480349344978166, + "grad_norm": 0.3239911145447236, + "learning_rate": 2.991540618266822e-05, + "loss": 0.2689, + "step": 3390 + }, + { + "epoch": 1.482532751091703, + "grad_norm": 0.31053885049245833, + "learning_rate": 2.9853117412903443e-05, + "loss": 0.259, + "step": 3395 + }, + { + "epoch": 1.48471615720524, + "grad_norm": 0.29558451827845905, + "learning_rate": 2.979079732024862e-05, + "loss": 0.2676, + "step": 3400 + }, + { + "epoch": 1.4868995633187774, + "grad_norm": 0.2919836637875955, + "learning_rate": 2.9728446306928813e-05, + "loss": 0.274, + "step": 3405 + }, + { + "epoch": 1.4890829694323144, + "grad_norm": 0.2914986515005281, + "learning_rate": 2.966606477536863e-05, + "loss": 0.265, + "step": 3410 + }, + { + "epoch": 1.4912663755458515, + "grad_norm": 0.24516343045152553, + "learning_rate": 2.960365312818967e-05, + "loss": 0.2745, + "step": 3415 + }, + { + "epoch": 1.4934497816593886, + "grad_norm": 0.31298311253304867, + "learning_rate": 2.9541211768207883e-05, + "loss": 0.2718, + "step": 3420 + }, + { + "epoch": 1.4956331877729259, + "grad_norm": 0.3182623875469872, + "learning_rate": 2.9478741098431027e-05, + "loss": 0.2747, + "step": 3425 + }, + { + "epoch": 1.497816593886463, + "grad_norm": 0.31623459423577244, + "learning_rate": 2.9416241522055987e-05, + "loss": 0.3007, + "step": 3430 + }, + { + "epoch": 1.5, + "grad_norm": 0.3659591897091837, + "learning_rate": 2.9353713442466246e-05, + "loss": 0.2797, + "step": 3435 + }, + { + "epoch": 1.502183406113537, + "grad_norm": 0.29892370752426095, + "learning_rate": 2.929115726322924e-05, + "loss": 0.2711, + "step": 3440 + }, + { + "epoch": 1.5043668122270741, + "grad_norm": 0.36468217436567685, + "learning_rate": 2.9228573388093765e-05, + "loss": 0.2781, + "step": 3445 + }, + { + "epoch": 1.5065502183406112, + "grad_norm": 0.36523975525457747, + "learning_rate": 2.916596222098738e-05, + "loss": 0.2782, + "step": 3450 + }, + { + "epoch": 1.5087336244541485, + "grad_norm": 0.24799351938184713, + "learning_rate": 2.9103324166013777e-05, + "loss": 0.2723, + "step": 3455 + }, + { + "epoch": 1.5109170305676856, + "grad_norm": 0.3395645662027844, + "learning_rate": 2.9040659627450195e-05, + "loss": 0.2734, + "step": 3460 + }, + { + "epoch": 1.5131004366812228, + "grad_norm": 0.2312365794117943, + "learning_rate": 2.897796900974481e-05, + "loss": 0.2695, + "step": 3465 + }, + { + "epoch": 1.51528384279476, + "grad_norm": 0.2733485343314042, + "learning_rate": 2.8915252717514108e-05, + "loss": 0.2753, + "step": 3470 + }, + { + "epoch": 1.517467248908297, + "grad_norm": 0.30853988506735613, + "learning_rate": 2.8852511155540286e-05, + "loss": 0.2805, + "step": 3475 + }, + { + "epoch": 1.519650655021834, + "grad_norm": 0.298316972177523, + "learning_rate": 2.8789744728768632e-05, + "loss": 0.2674, + "step": 3480 + }, + { + "epoch": 1.5218340611353711, + "grad_norm": 0.33489826338103257, + "learning_rate": 2.8726953842304917e-05, + "loss": 0.2781, + "step": 3485 + }, + { + "epoch": 1.5240174672489082, + "grad_norm": 0.29180816733941406, + "learning_rate": 2.8664138901412784e-05, + "loss": 0.2828, + "step": 3490 + }, + { + "epoch": 1.5262008733624453, + "grad_norm": 0.2652928534046041, + "learning_rate": 2.8601300311511135e-05, + "loss": 0.2704, + "step": 3495 + }, + { + "epoch": 1.5283842794759825, + "grad_norm": 0.2478372696723595, + "learning_rate": 2.8538438478171493e-05, + "loss": 0.2709, + "step": 3500 + }, + { + "epoch": 1.5305676855895196, + "grad_norm": 0.29590356196390866, + "learning_rate": 2.8475553807115392e-05, + "loss": 0.2645, + "step": 3505 + }, + { + "epoch": 1.532751091703057, + "grad_norm": 0.2404106523342164, + "learning_rate": 2.8412646704211787e-05, + "loss": 0.2716, + "step": 3510 + }, + { + "epoch": 1.534934497816594, + "grad_norm": 0.2913911540734402, + "learning_rate": 2.8349717575474395e-05, + "loss": 0.2812, + "step": 3515 + }, + { + "epoch": 1.537117903930131, + "grad_norm": 0.27619563545275305, + "learning_rate": 2.8286766827059103e-05, + "loss": 0.2852, + "step": 3520 + }, + { + "epoch": 1.539301310043668, + "grad_norm": 0.25296146919214546, + "learning_rate": 2.8223794865261326e-05, + "loss": 0.2706, + "step": 3525 + }, + { + "epoch": 1.5414847161572052, + "grad_norm": 0.4120208095199771, + "learning_rate": 2.8160802096513396e-05, + "loss": 0.2655, + "step": 3530 + }, + { + "epoch": 1.5436681222707422, + "grad_norm": 0.2797662392218346, + "learning_rate": 2.809778892738194e-05, + "loss": 0.2799, + "step": 3535 + }, + { + "epoch": 1.5458515283842795, + "grad_norm": 0.30491897219029096, + "learning_rate": 2.8034755764565252e-05, + "loss": 0.2703, + "step": 3540 + }, + { + "epoch": 1.5480349344978166, + "grad_norm": 0.2830893746174212, + "learning_rate": 2.797170301489066e-05, + "loss": 0.2716, + "step": 3545 + }, + { + "epoch": 1.5502183406113537, + "grad_norm": 0.23485142267745251, + "learning_rate": 2.7908631085311933e-05, + "loss": 0.2617, + "step": 3550 + }, + { + "epoch": 1.552401746724891, + "grad_norm": 0.3021327029538882, + "learning_rate": 2.78455403829066e-05, + "loss": 0.2638, + "step": 3555 + }, + { + "epoch": 1.554585152838428, + "grad_norm": 0.2621866662981608, + "learning_rate": 2.7782431314873377e-05, + "loss": 0.2678, + "step": 3560 + }, + { + "epoch": 1.556768558951965, + "grad_norm": 0.2747099055332518, + "learning_rate": 2.7719304288529503e-05, + "loss": 0.2714, + "step": 3565 + }, + { + "epoch": 1.5589519650655022, + "grad_norm": 0.27602610129495325, + "learning_rate": 2.7656159711308127e-05, + "loss": 0.2754, + "step": 3570 + }, + { + "epoch": 1.5611353711790392, + "grad_norm": 0.2563630850581704, + "learning_rate": 2.7592997990755665e-05, + "loss": 0.2674, + "step": 3575 + }, + { + "epoch": 1.5633187772925763, + "grad_norm": 0.25276036483666414, + "learning_rate": 2.75298195345292e-05, + "loss": 0.2689, + "step": 3580 + }, + { + "epoch": 1.5655021834061136, + "grad_norm": 0.2568576656221897, + "learning_rate": 2.7466624750393804e-05, + "loss": 0.2624, + "step": 3585 + }, + { + "epoch": 1.5676855895196506, + "grad_norm": 0.2540380161679821, + "learning_rate": 2.740341404621996e-05, + "loss": 0.2704, + "step": 3590 + }, + { + "epoch": 1.569868995633188, + "grad_norm": 0.2691098927474828, + "learning_rate": 2.7340187829980884e-05, + "loss": 0.2596, + "step": 3595 + }, + { + "epoch": 1.572052401746725, + "grad_norm": 0.27708252166503045, + "learning_rate": 2.7276946509749905e-05, + "loss": 0.2753, + "step": 3600 + }, + { + "epoch": 1.574235807860262, + "grad_norm": 0.2828335584788292, + "learning_rate": 2.7213690493697852e-05, + "loss": 0.2813, + "step": 3605 + }, + { + "epoch": 1.5764192139737991, + "grad_norm": 0.29157807173338984, + "learning_rate": 2.7150420190090397e-05, + "loss": 0.2654, + "step": 3610 + }, + { + "epoch": 1.5786026200873362, + "grad_norm": 0.2621619308124079, + "learning_rate": 2.7087136007285417e-05, + "loss": 0.269, + "step": 3615 + }, + { + "epoch": 1.5807860262008733, + "grad_norm": 0.27055849925403364, + "learning_rate": 2.7023838353730386e-05, + "loss": 0.2603, + "step": 3620 + }, + { + "epoch": 1.5829694323144103, + "grad_norm": 0.28812667281668847, + "learning_rate": 2.6960527637959704e-05, + "loss": 0.2615, + "step": 3625 + }, + { + "epoch": 1.5851528384279476, + "grad_norm": 0.3648305241785186, + "learning_rate": 2.689720426859208e-05, + "loss": 0.2745, + "step": 3630 + }, + { + "epoch": 1.5873362445414847, + "grad_norm": 0.3579531717893759, + "learning_rate": 2.68338686543279e-05, + "loss": 0.2754, + "step": 3635 + }, + { + "epoch": 1.589519650655022, + "grad_norm": 0.30496840176534284, + "learning_rate": 2.6770521203946582e-05, + "loss": 0.2687, + "step": 3640 + }, + { + "epoch": 1.591703056768559, + "grad_norm": 0.3316407097346181, + "learning_rate": 2.670716232630392e-05, + "loss": 0.2708, + "step": 3645 + }, + { + "epoch": 1.5938864628820961, + "grad_norm": 0.30884262120013556, + "learning_rate": 2.6643792430329473e-05, + "loss": 0.2691, + "step": 3650 + }, + { + "epoch": 1.5960698689956332, + "grad_norm": 0.24575164955804643, + "learning_rate": 2.6580411925023916e-05, + "loss": 0.2895, + "step": 3655 + }, + { + "epoch": 1.5982532751091703, + "grad_norm": 0.2403162634268798, + "learning_rate": 2.6517021219456402e-05, + "loss": 0.2768, + "step": 3660 + }, + { + "epoch": 1.6004366812227073, + "grad_norm": 0.7286289540628894, + "learning_rate": 2.6453620722761896e-05, + "loss": 0.2772, + "step": 3665 + }, + { + "epoch": 1.6026200873362444, + "grad_norm": 0.2698469031360812, + "learning_rate": 2.6390210844138593e-05, + "loss": 0.2772, + "step": 3670 + }, + { + "epoch": 1.6048034934497817, + "grad_norm": 0.29594772177235845, + "learning_rate": 2.6326791992845202e-05, + "loss": 0.2746, + "step": 3675 + }, + { + "epoch": 1.6069868995633187, + "grad_norm": 0.3432286485907405, + "learning_rate": 2.626336457819837e-05, + "loss": 0.2627, + "step": 3680 + }, + { + "epoch": 1.609170305676856, + "grad_norm": 0.47848404072485456, + "learning_rate": 2.619992900957e-05, + "loss": 0.2896, + "step": 3685 + }, + { + "epoch": 1.611353711790393, + "grad_norm": 0.400345398790263, + "learning_rate": 2.613648569638464e-05, + "loss": 0.2832, + "step": 3690 + }, + { + "epoch": 1.6135371179039302, + "grad_norm": 0.26168308925039074, + "learning_rate": 2.6073035048116796e-05, + "loss": 0.2693, + "step": 3695 + }, + { + "epoch": 1.6157205240174672, + "grad_norm": 0.33192980592740756, + "learning_rate": 2.6009577474288333e-05, + "loss": 0.2759, + "step": 3700 + }, + { + "epoch": 1.6179039301310043, + "grad_norm": 0.2787526637545283, + "learning_rate": 2.5946113384465813e-05, + "loss": 0.2642, + "step": 3705 + }, + { + "epoch": 1.6200873362445414, + "grad_norm": 0.27338870781942926, + "learning_rate": 2.5882643188257856e-05, + "loss": 0.2706, + "step": 3710 + }, + { + "epoch": 1.6222707423580784, + "grad_norm": 0.2435797974873236, + "learning_rate": 2.5819167295312485e-05, + "loss": 0.2724, + "step": 3715 + }, + { + "epoch": 1.6244541484716157, + "grad_norm": 0.29781182108189724, + "learning_rate": 2.57556861153145e-05, + "loss": 0.2707, + "step": 3720 + }, + { + "epoch": 1.6266375545851528, + "grad_norm": 0.32478484905425753, + "learning_rate": 2.569220005798281e-05, + "loss": 0.2725, + "step": 3725 + }, + { + "epoch": 1.62882096069869, + "grad_norm": 0.2669336390561608, + "learning_rate": 2.5628709533067824e-05, + "loss": 0.2713, + "step": 3730 + }, + { + "epoch": 1.6310043668122272, + "grad_norm": 0.2573882639282421, + "learning_rate": 2.556521495034877e-05, + "loss": 0.2606, + "step": 3735 + }, + { + "epoch": 1.6331877729257642, + "grad_norm": 0.2705880029160448, + "learning_rate": 2.550171671963108e-05, + "loss": 0.2575, + "step": 3740 + }, + { + "epoch": 1.6353711790393013, + "grad_norm": 0.27796104371371233, + "learning_rate": 2.543821525074371e-05, + "loss": 0.2699, + "step": 3745 + }, + { + "epoch": 1.6375545851528384, + "grad_norm": 0.2968772131918911, + "learning_rate": 2.537471095353653e-05, + "loss": 0.2604, + "step": 3750 + }, + { + "epoch": 1.6397379912663754, + "grad_norm": 0.30473359891717344, + "learning_rate": 2.5311204237877657e-05, + "loss": 0.2715, + "step": 3755 + }, + { + "epoch": 1.6419213973799127, + "grad_norm": 0.29986937512458306, + "learning_rate": 2.524769551365083e-05, + "loss": 0.2569, + "step": 3760 + }, + { + "epoch": 1.6441048034934498, + "grad_norm": 0.29756570977438257, + "learning_rate": 2.5184185190752752e-05, + "loss": 0.2805, + "step": 3765 + }, + { + "epoch": 1.6462882096069869, + "grad_norm": 0.24764879064376663, + "learning_rate": 2.512067367909041e-05, + "loss": 0.2723, + "step": 3770 + }, + { + "epoch": 1.6484716157205241, + "grad_norm": 0.2536796536666786, + "learning_rate": 2.505716138857851e-05, + "loss": 0.2692, + "step": 3775 + }, + { + "epoch": 1.6506550218340612, + "grad_norm": 0.27985489044371364, + "learning_rate": 2.499364872913676e-05, + "loss": 0.2736, + "step": 3780 + }, + { + "epoch": 1.6528384279475983, + "grad_norm": 0.24159737014671576, + "learning_rate": 2.4930136110687246e-05, + "loss": 0.2711, + "step": 3785 + }, + { + "epoch": 1.6550218340611353, + "grad_norm": 0.28234624412572884, + "learning_rate": 2.486662394315181e-05, + "loss": 0.2865, + "step": 3790 + }, + { + "epoch": 1.6572052401746724, + "grad_norm": 0.2491348077927823, + "learning_rate": 2.4803112636449355e-05, + "loss": 0.2731, + "step": 3795 + }, + { + "epoch": 1.6593886462882095, + "grad_norm": 0.31780593847271627, + "learning_rate": 2.473960260049325e-05, + "loss": 0.2789, + "step": 3800 + }, + { + "epoch": 1.6615720524017468, + "grad_norm": 0.2685872565371744, + "learning_rate": 2.4676094245188658e-05, + "loss": 0.2933, + "step": 3805 + }, + { + "epoch": 1.6637554585152838, + "grad_norm": 0.263863501194976, + "learning_rate": 2.461258798042989e-05, + "loss": 0.2685, + "step": 3810 + }, + { + "epoch": 1.6659388646288211, + "grad_norm": 0.21766584856793741, + "learning_rate": 2.454908421609777e-05, + "loss": 0.2552, + "step": 3815 + }, + { + "epoch": 1.6681222707423582, + "grad_norm": 0.27428707449471423, + "learning_rate": 2.4485583362056974e-05, + "loss": 0.2786, + "step": 3820 + }, + { + "epoch": 1.6703056768558953, + "grad_norm": 0.2835969591758734, + "learning_rate": 2.4422085828153403e-05, + "loss": 0.2662, + "step": 3825 + }, + { + "epoch": 1.6724890829694323, + "grad_norm": 0.29526168010306597, + "learning_rate": 2.435859202421153e-05, + "loss": 0.269, + "step": 3830 + }, + { + "epoch": 1.6746724890829694, + "grad_norm": 0.30337514206743876, + "learning_rate": 2.429510236003175e-05, + "loss": 0.2597, + "step": 3835 + }, + { + "epoch": 1.6768558951965065, + "grad_norm": 0.2584878592869018, + "learning_rate": 2.4231617245387745e-05, + "loss": 0.2709, + "step": 3840 + }, + { + "epoch": 1.6790393013100435, + "grad_norm": 0.27852560485634303, + "learning_rate": 2.416813709002382e-05, + "loss": 0.2712, + "step": 3845 + }, + { + "epoch": 1.6812227074235808, + "grad_norm": 0.31941354537015787, + "learning_rate": 2.4104662303652288e-05, + "loss": 0.286, + "step": 3850 + }, + { + "epoch": 1.6834061135371179, + "grad_norm": 0.2639262509863191, + "learning_rate": 2.4041193295950788e-05, + "loss": 0.2679, + "step": 3855 + }, + { + "epoch": 1.6855895196506552, + "grad_norm": 0.27648769301544285, + "learning_rate": 2.39777304765597e-05, + "loss": 0.2764, + "step": 3860 + }, + { + "epoch": 1.6877729257641922, + "grad_norm": 0.2557231698124579, + "learning_rate": 2.391427425507943e-05, + "loss": 0.2773, + "step": 3865 + }, + { + "epoch": 1.6899563318777293, + "grad_norm": 0.2835965272930869, + "learning_rate": 2.385082504106781e-05, + "loss": 0.2655, + "step": 3870 + }, + { + "epoch": 1.6921397379912664, + "grad_norm": 0.30219688777437337, + "learning_rate": 2.378738324403744e-05, + "loss": 0.2649, + "step": 3875 + }, + { + "epoch": 1.6943231441048034, + "grad_norm": 0.286618784377373, + "learning_rate": 2.3723949273453073e-05, + "loss": 0.2668, + "step": 3880 + }, + { + "epoch": 1.6965065502183405, + "grad_norm": 0.31422639752909204, + "learning_rate": 2.3660523538728904e-05, + "loss": 0.2671, + "step": 3885 + }, + { + "epoch": 1.6986899563318776, + "grad_norm": 0.24899445840743908, + "learning_rate": 2.3597106449226015e-05, + "loss": 0.2603, + "step": 3890 + }, + { + "epoch": 1.7008733624454149, + "grad_norm": 0.28778299708675603, + "learning_rate": 2.3533698414249668e-05, + "loss": 0.2532, + "step": 3895 + }, + { + "epoch": 1.703056768558952, + "grad_norm": 0.3010236993612902, + "learning_rate": 2.3470299843046693e-05, + "loss": 0.2607, + "step": 3900 + }, + { + "epoch": 1.7052401746724892, + "grad_norm": 0.2505565783956947, + "learning_rate": 2.340691114480284e-05, + "loss": 0.2675, + "step": 3905 + }, + { + "epoch": 1.7074235807860263, + "grad_norm": 0.2534199791327587, + "learning_rate": 2.334353272864013e-05, + "loss": 0.2727, + "step": 3910 + }, + { + "epoch": 1.7096069868995634, + "grad_norm": 0.2759471253942165, + "learning_rate": 2.3280165003614227e-05, + "loss": 0.2663, + "step": 3915 + }, + { + "epoch": 1.7117903930131004, + "grad_norm": 0.2958147406172001, + "learning_rate": 2.3216808378711788e-05, + "loss": 0.2705, + "step": 3920 + }, + { + "epoch": 1.7139737991266375, + "grad_norm": 0.26156474466311214, + "learning_rate": 2.3153463262847835e-05, + "loss": 0.242, + "step": 3925 + }, + { + "epoch": 1.7161572052401746, + "grad_norm": 0.29968058749209264, + "learning_rate": 2.3090130064863113e-05, + "loss": 0.2681, + "step": 3930 + }, + { + "epoch": 1.7183406113537116, + "grad_norm": 0.24513602810052307, + "learning_rate": 2.3026809193521423e-05, + "loss": 0.2746, + "step": 3935 + }, + { + "epoch": 1.720524017467249, + "grad_norm": 0.2624738237172384, + "learning_rate": 2.296350105750703e-05, + "loss": 0.2682, + "step": 3940 + }, + { + "epoch": 1.722707423580786, + "grad_norm": 0.21915397316253904, + "learning_rate": 2.2900206065421997e-05, + "loss": 0.26, + "step": 3945 + }, + { + "epoch": 1.7248908296943233, + "grad_norm": 0.248859562817194, + "learning_rate": 2.2836924625783554e-05, + "loss": 0.2776, + "step": 3950 + }, + { + "epoch": 1.7270742358078603, + "grad_norm": 0.2941037210948601, + "learning_rate": 2.2773657147021464e-05, + "loss": 0.281, + "step": 3955 + }, + { + "epoch": 1.7292576419213974, + "grad_norm": 0.22756003553556517, + "learning_rate": 2.271040403747537e-05, + "loss": 0.2597, + "step": 3960 + }, + { + "epoch": 1.7314410480349345, + "grad_norm": 0.274949859317834, + "learning_rate": 2.2647165705392186e-05, + "loss": 0.2697, + "step": 3965 + }, + { + "epoch": 1.7336244541484715, + "grad_norm": 0.2797174304909422, + "learning_rate": 2.2583942558923456e-05, + "loss": 0.2716, + "step": 3970 + }, + { + "epoch": 1.7358078602620086, + "grad_norm": 0.26704187553881054, + "learning_rate": 2.2520735006122688e-05, + "loss": 0.274, + "step": 3975 + }, + { + "epoch": 1.737991266375546, + "grad_norm": 0.284507084979283, + "learning_rate": 2.2457543454942786e-05, + "loss": 0.2587, + "step": 3980 + }, + { + "epoch": 1.740174672489083, + "grad_norm": 0.2699117094732979, + "learning_rate": 2.2394368313233332e-05, + "loss": 0.2619, + "step": 3985 + }, + { + "epoch": 1.74235807860262, + "grad_norm": 0.25250488525353865, + "learning_rate": 2.2331209988738024e-05, + "loss": 0.2602, + "step": 3990 + }, + { + "epoch": 1.7445414847161573, + "grad_norm": 0.24206728711747397, + "learning_rate": 2.226806888909202e-05, + "loss": 0.262, + "step": 3995 + }, + { + "epoch": 1.7467248908296944, + "grad_norm": 0.2907832129799141, + "learning_rate": 2.2204945421819292e-05, + "loss": 0.2636, + "step": 4000 + }, + { + "epoch": 1.7489082969432315, + "grad_norm": 0.29699912162411024, + "learning_rate": 2.2141839994330027e-05, + "loss": 0.2747, + "step": 4005 + }, + { + "epoch": 1.7510917030567685, + "grad_norm": 0.2505983156960274, + "learning_rate": 2.207875301391796e-05, + "loss": 0.2706, + "step": 4010 + }, + { + "epoch": 1.7532751091703056, + "grad_norm": 0.20887238006814582, + "learning_rate": 2.2015684887757776e-05, + "loss": 0.2544, + "step": 4015 + }, + { + "epoch": 1.7554585152838427, + "grad_norm": 0.2350828605757233, + "learning_rate": 2.195263602290248e-05, + "loss": 0.2759, + "step": 4020 + }, + { + "epoch": 1.75764192139738, + "grad_norm": 0.27919466794671643, + "learning_rate": 2.188960682628074e-05, + "loss": 0.2468, + "step": 4025 + }, + { + "epoch": 1.759825327510917, + "grad_norm": 0.24755062875053999, + "learning_rate": 2.1826597704694306e-05, + "loss": 0.2699, + "step": 4030 + }, + { + "epoch": 1.762008733624454, + "grad_norm": 0.28930040508547633, + "learning_rate": 2.176360906481533e-05, + "loss": 0.27, + "step": 4035 + }, + { + "epoch": 1.7641921397379914, + "grad_norm": 0.2463760850445127, + "learning_rate": 2.1700641313183807e-05, + "loss": 0.2604, + "step": 4040 + }, + { + "epoch": 1.7663755458515285, + "grad_norm": 0.24878427896848646, + "learning_rate": 2.163769485620489e-05, + "loss": 0.2647, + "step": 4045 + }, + { + "epoch": 1.7685589519650655, + "grad_norm": 0.3491349213382724, + "learning_rate": 2.1574770100146295e-05, + "loss": 0.2717, + "step": 4050 + }, + { + "epoch": 1.7707423580786026, + "grad_norm": 0.26013086945480984, + "learning_rate": 2.1511867451135693e-05, + "loss": 0.2599, + "step": 4055 + }, + { + "epoch": 1.7729257641921397, + "grad_norm": 0.2900875329162927, + "learning_rate": 2.1448987315158048e-05, + "loss": 0.2686, + "step": 4060 + }, + { + "epoch": 1.7751091703056767, + "grad_norm": 0.2653600164327816, + "learning_rate": 2.1386130098053036e-05, + "loss": 0.2627, + "step": 4065 + }, + { + "epoch": 1.777292576419214, + "grad_norm": 0.27303311100886735, + "learning_rate": 2.1323296205512418e-05, + "loss": 0.2708, + "step": 4070 + }, + { + "epoch": 1.779475982532751, + "grad_norm": 0.29210046654985766, + "learning_rate": 2.1260486043077388e-05, + "loss": 0.2581, + "step": 4075 + }, + { + "epoch": 1.7816593886462884, + "grad_norm": 0.284149374163666, + "learning_rate": 2.1197700016136014e-05, + "loss": 0.266, + "step": 4080 + }, + { + "epoch": 1.7838427947598254, + "grad_norm": 0.2409781119667348, + "learning_rate": 2.113493852992056e-05, + "loss": 0.2571, + "step": 4085 + }, + { + "epoch": 1.7860262008733625, + "grad_norm": 0.20843436142983807, + "learning_rate": 2.1072201989504913e-05, + "loss": 0.2733, + "step": 4090 + }, + { + "epoch": 1.7882096069868996, + "grad_norm": 0.23230387555929172, + "learning_rate": 2.1009490799801965e-05, + "loss": 0.269, + "step": 4095 + }, + { + "epoch": 1.7903930131004366, + "grad_norm": 0.2694983772096665, + "learning_rate": 2.0946805365560966e-05, + "loss": 0.2745, + "step": 4100 + }, + { + "epoch": 1.7925764192139737, + "grad_norm": 0.24747448881142187, + "learning_rate": 2.0884146091364956e-05, + "loss": 0.2659, + "step": 4105 + }, + { + "epoch": 1.7947598253275108, + "grad_norm": 0.22323270338570358, + "learning_rate": 2.0821513381628132e-05, + "loss": 0.2645, + "step": 4110 + }, + { + "epoch": 1.796943231441048, + "grad_norm": 0.2580592437636737, + "learning_rate": 2.0758907640593228e-05, + "loss": 0.2563, + "step": 4115 + }, + { + "epoch": 1.7991266375545851, + "grad_norm": 0.2411641382308118, + "learning_rate": 2.0696329272328932e-05, + "loss": 0.2721, + "step": 4120 + }, + { + "epoch": 1.8013100436681224, + "grad_norm": 0.25806558253779854, + "learning_rate": 2.063377868072724e-05, + "loss": 0.2616, + "step": 4125 + }, + { + "epoch": 1.8034934497816595, + "grad_norm": 0.2455964016028092, + "learning_rate": 2.0571256269500893e-05, + "loss": 0.271, + "step": 4130 + }, + { + "epoch": 1.8056768558951966, + "grad_norm": 0.25469457274681495, + "learning_rate": 2.0508762442180745e-05, + "loss": 0.2683, + "step": 4135 + }, + { + "epoch": 1.8078602620087336, + "grad_norm": 0.2734524817312364, + "learning_rate": 2.0446297602113167e-05, + "loss": 0.2782, + "step": 4140 + }, + { + "epoch": 1.8100436681222707, + "grad_norm": 0.2711480537954455, + "learning_rate": 2.0383862152457432e-05, + "loss": 0.2636, + "step": 4145 + }, + { + "epoch": 1.8122270742358078, + "grad_norm": 0.23192737600539587, + "learning_rate": 2.0321456496183134e-05, + "loss": 0.276, + "step": 4150 + }, + { + "epoch": 1.8144104803493448, + "grad_norm": 0.2610658805286943, + "learning_rate": 2.0259081036067563e-05, + "loss": 0.2706, + "step": 4155 + }, + { + "epoch": 1.8165938864628821, + "grad_norm": 0.2670079724190112, + "learning_rate": 2.0196736174693126e-05, + "loss": 0.2529, + "step": 4160 + }, + { + "epoch": 1.8187772925764192, + "grad_norm": 0.25454455186632396, + "learning_rate": 2.013442231444474e-05, + "loss": 0.2649, + "step": 4165 + }, + { + "epoch": 1.8209606986899565, + "grad_norm": 0.2777703715553109, + "learning_rate": 2.007213985750724e-05, + "loss": 0.265, + "step": 4170 + }, + { + "epoch": 1.8231441048034935, + "grad_norm": 0.2654974710451674, + "learning_rate": 2.000988920586275e-05, + "loss": 0.2653, + "step": 4175 + }, + { + "epoch": 1.8253275109170306, + "grad_norm": 0.28984168869856813, + "learning_rate": 1.994767076128816e-05, + "loss": 0.2572, + "step": 4180 + }, + { + "epoch": 1.8275109170305677, + "grad_norm": 0.2620326885283614, + "learning_rate": 1.988548492535246e-05, + "loss": 0.2754, + "step": 4185 + }, + { + "epoch": 1.8296943231441047, + "grad_norm": 0.24235633152361688, + "learning_rate": 1.9823332099414185e-05, + "loss": 0.2717, + "step": 4190 + }, + { + "epoch": 1.8318777292576418, + "grad_norm": 0.24991695967334396, + "learning_rate": 1.9761212684618824e-05, + "loss": 0.25, + "step": 4195 + }, + { + "epoch": 1.8340611353711789, + "grad_norm": 0.21335483219532675, + "learning_rate": 1.9699127081896214e-05, + "loss": 0.2513, + "step": 4200 + }, + { + "epoch": 1.8362445414847162, + "grad_norm": 0.2515339708285586, + "learning_rate": 1.963707569195797e-05, + "loss": 0.2524, + "step": 4205 + }, + { + "epoch": 1.8384279475982532, + "grad_norm": 0.23802145795043225, + "learning_rate": 1.9575058915294885e-05, + "loss": 0.2652, + "step": 4210 + }, + { + "epoch": 1.8406113537117905, + "grad_norm": 0.30056095101017133, + "learning_rate": 1.9513077152174363e-05, + "loss": 0.2776, + "step": 4215 + }, + { + "epoch": 1.8427947598253276, + "grad_norm": 0.2915453674361558, + "learning_rate": 1.9451130802637822e-05, + "loss": 0.2588, + "step": 4220 + }, + { + "epoch": 1.8449781659388647, + "grad_norm": 0.30777863799002625, + "learning_rate": 1.9389220266498102e-05, + "loss": 0.2701, + "step": 4225 + }, + { + "epoch": 1.8471615720524017, + "grad_norm": 0.2597777533142635, + "learning_rate": 1.9327345943336904e-05, + "loss": 0.2645, + "step": 4230 + }, + { + "epoch": 1.8493449781659388, + "grad_norm": 0.3080444941319507, + "learning_rate": 1.9265508232502208e-05, + "loss": 0.2749, + "step": 4235 + }, + { + "epoch": 1.8515283842794759, + "grad_norm": 0.2831226100931665, + "learning_rate": 1.9203707533105687e-05, + "loss": 0.2598, + "step": 4240 + }, + { + "epoch": 1.8537117903930131, + "grad_norm": 0.27124968567558605, + "learning_rate": 1.914194424402015e-05, + "loss": 0.271, + "step": 4245 + }, + { + "epoch": 1.8558951965065502, + "grad_norm": 0.24173431823915756, + "learning_rate": 1.9080218763876917e-05, + "loss": 0.2512, + "step": 4250 + }, + { + "epoch": 1.8580786026200873, + "grad_norm": 0.27847599015903607, + "learning_rate": 1.901853149106332e-05, + "loss": 0.263, + "step": 4255 + }, + { + "epoch": 1.8602620087336246, + "grad_norm": 0.348403165189276, + "learning_rate": 1.8956882823720075e-05, + "loss": 0.2657, + "step": 4260 + }, + { + "epoch": 1.8624454148471616, + "grad_norm": 0.24233816768582403, + "learning_rate": 1.889527315973874e-05, + "loss": 0.26, + "step": 4265 + }, + { + "epoch": 1.8646288209606987, + "grad_norm": 0.23840655638016398, + "learning_rate": 1.8833702896759132e-05, + "loss": 0.2636, + "step": 4270 + }, + { + "epoch": 1.8668122270742358, + "grad_norm": 0.2549222906252635, + "learning_rate": 1.8772172432166767e-05, + "loss": 0.2744, + "step": 4275 + }, + { + "epoch": 1.8689956331877728, + "grad_norm": 0.25475120516780736, + "learning_rate": 1.87106821630903e-05, + "loss": 0.2748, + "step": 4280 + }, + { + "epoch": 1.87117903930131, + "grad_norm": 0.24088260430162353, + "learning_rate": 1.8649232486398955e-05, + "loss": 0.2638, + "step": 4285 + }, + { + "epoch": 1.8733624454148472, + "grad_norm": 0.2556026511315039, + "learning_rate": 1.8587823798699958e-05, + "loss": 0.2601, + "step": 4290 + }, + { + "epoch": 1.8755458515283843, + "grad_norm": 0.2700718784491458, + "learning_rate": 1.8526456496335997e-05, + "loss": 0.2525, + "step": 4295 + }, + { + "epoch": 1.8777292576419216, + "grad_norm": 0.27667715242466545, + "learning_rate": 1.846513097538265e-05, + "loss": 0.2547, + "step": 4300 + }, + { + "epoch": 1.8799126637554586, + "grad_norm": 0.2488594588357838, + "learning_rate": 1.8403847631645815e-05, + "loss": 0.2747, + "step": 4305 + }, + { + "epoch": 1.8820960698689957, + "grad_norm": 0.2486103000672603, + "learning_rate": 1.834260686065919e-05, + "loss": 0.2755, + "step": 4310 + }, + { + "epoch": 1.8842794759825328, + "grad_norm": 0.2673003900981902, + "learning_rate": 1.8281409057681685e-05, + "loss": 0.2556, + "step": 4315 + }, + { + "epoch": 1.8864628820960698, + "grad_norm": 0.27816423985109073, + "learning_rate": 1.8220254617694904e-05, + "loss": 0.2725, + "step": 4320 + }, + { + "epoch": 1.888646288209607, + "grad_norm": 0.21891089387179588, + "learning_rate": 1.815914393540057e-05, + "loss": 0.2672, + "step": 4325 + }, + { + "epoch": 1.890829694323144, + "grad_norm": 0.3027560989030244, + "learning_rate": 1.8098077405217973e-05, + "loss": 0.2681, + "step": 4330 + }, + { + "epoch": 1.8930131004366813, + "grad_norm": 0.2956885379322635, + "learning_rate": 1.8037055421281453e-05, + "loss": 0.2754, + "step": 4335 + }, + { + "epoch": 1.8951965065502183, + "grad_norm": 0.27344213459308064, + "learning_rate": 1.7976078377437843e-05, + "loss": 0.254, + "step": 4340 + }, + { + "epoch": 1.8973799126637556, + "grad_norm": 0.25908804251433426, + "learning_rate": 1.7915146667243912e-05, + "loss": 0.2709, + "step": 4345 + }, + { + "epoch": 1.8995633187772927, + "grad_norm": 0.27387774174550766, + "learning_rate": 1.785426068396385e-05, + "loss": 0.2724, + "step": 4350 + }, + { + "epoch": 1.9017467248908297, + "grad_norm": 0.2349626027532333, + "learning_rate": 1.7793420820566698e-05, + "loss": 0.2685, + "step": 4355 + }, + { + "epoch": 1.9039301310043668, + "grad_norm": 0.27519209684053136, + "learning_rate": 1.7732627469723867e-05, + "loss": 0.2665, + "step": 4360 + }, + { + "epoch": 1.9061135371179039, + "grad_norm": 0.2237579472013528, + "learning_rate": 1.7671881023806526e-05, + "loss": 0.2629, + "step": 4365 + }, + { + "epoch": 1.908296943231441, + "grad_norm": 0.24463587078424295, + "learning_rate": 1.7611181874883133e-05, + "loss": 0.2637, + "step": 4370 + }, + { + "epoch": 1.910480349344978, + "grad_norm": 0.30256331204318965, + "learning_rate": 1.755053041471689e-05, + "loss": 0.2626, + "step": 4375 + }, + { + "epoch": 1.9126637554585153, + "grad_norm": 0.23702938555275288, + "learning_rate": 1.748992703476318e-05, + "loss": 0.2621, + "step": 4380 + }, + { + "epoch": 1.9148471615720524, + "grad_norm": 0.29244916287768213, + "learning_rate": 1.7429372126167105e-05, + "loss": 0.2492, + "step": 4385 + }, + { + "epoch": 1.9170305676855897, + "grad_norm": 0.2588931757550786, + "learning_rate": 1.7368866079760887e-05, + "loss": 0.2615, + "step": 4390 + }, + { + "epoch": 1.9192139737991267, + "grad_norm": 0.25196463301087313, + "learning_rate": 1.7308409286061407e-05, + "loss": 0.2686, + "step": 4395 + }, + { + "epoch": 1.9213973799126638, + "grad_norm": 0.2158443773081268, + "learning_rate": 1.7248002135267645e-05, + "loss": 0.2666, + "step": 4400 + }, + { + "epoch": 1.9235807860262009, + "grad_norm": 0.287220788904416, + "learning_rate": 1.7187645017258196e-05, + "loss": 0.2735, + "step": 4405 + }, + { + "epoch": 1.925764192139738, + "grad_norm": 0.23107125712315182, + "learning_rate": 1.712733832158872e-05, + "loss": 0.2613, + "step": 4410 + }, + { + "epoch": 1.927947598253275, + "grad_norm": 0.24853969589485442, + "learning_rate": 1.7067082437489436e-05, + "loss": 0.2653, + "step": 4415 + }, + { + "epoch": 1.930131004366812, + "grad_norm": 0.2767451923179264, + "learning_rate": 1.7006877753862632e-05, + "loss": 0.262, + "step": 4420 + }, + { + "epoch": 1.9323144104803494, + "grad_norm": 0.22927033548053927, + "learning_rate": 1.6946724659280133e-05, + "loss": 0.2542, + "step": 4425 + }, + { + "epoch": 1.9344978165938864, + "grad_norm": 0.23110650040140782, + "learning_rate": 1.6886623541980796e-05, + "loss": 0.2757, + "step": 4430 + }, + { + "epoch": 1.9366812227074237, + "grad_norm": 0.23731032145161848, + "learning_rate": 1.682657478986801e-05, + "loss": 0.2645, + "step": 4435 + }, + { + "epoch": 1.9388646288209608, + "grad_norm": 0.25552976383740716, + "learning_rate": 1.6766578790507192e-05, + "loss": 0.2595, + "step": 4440 + }, + { + "epoch": 1.9410480349344978, + "grad_norm": 0.2155945620763327, + "learning_rate": 1.6706635931123277e-05, + "loss": 0.2596, + "step": 4445 + }, + { + "epoch": 1.943231441048035, + "grad_norm": 0.24247256828652752, + "learning_rate": 1.664674659859822e-05, + "loss": 0.2493, + "step": 4450 + }, + { + "epoch": 1.945414847161572, + "grad_norm": 0.2516593577449982, + "learning_rate": 1.658691117946853e-05, + "loss": 0.2761, + "step": 4455 + }, + { + "epoch": 1.947598253275109, + "grad_norm": 0.21991257872575132, + "learning_rate": 1.6527130059922717e-05, + "loss": 0.2632, + "step": 4460 + }, + { + "epoch": 1.9497816593886463, + "grad_norm": 0.261259312013636, + "learning_rate": 1.646740362579886e-05, + "loss": 0.2615, + "step": 4465 + }, + { + "epoch": 1.9519650655021834, + "grad_norm": 0.2318715733570351, + "learning_rate": 1.6407732262582067e-05, + "loss": 0.2654, + "step": 4470 + }, + { + "epoch": 1.9541484716157205, + "grad_norm": 0.22989633286543176, + "learning_rate": 1.634811635540201e-05, + "loss": 0.2593, + "step": 4475 + }, + { + "epoch": 1.9563318777292578, + "grad_norm": 0.24017270832452148, + "learning_rate": 1.6288556289030452e-05, + "loss": 0.2676, + "step": 4480 + }, + { + "epoch": 1.9585152838427948, + "grad_norm": 0.25142686822052845, + "learning_rate": 1.6229052447878743e-05, + "loss": 0.2541, + "step": 4485 + }, + { + "epoch": 1.960698689956332, + "grad_norm": 0.22732754650526263, + "learning_rate": 1.6169605215995345e-05, + "loss": 0.259, + "step": 4490 + }, + { + "epoch": 1.962882096069869, + "grad_norm": 0.24853011047008652, + "learning_rate": 1.6110214977063345e-05, + "loss": 0.2559, + "step": 4495 + }, + { + "epoch": 1.965065502183406, + "grad_norm": 0.2557640948437553, + "learning_rate": 1.6050882114398e-05, + "loss": 0.2508, + "step": 4500 + }, + { + "epoch": 1.967248908296943, + "grad_norm": 0.24324444806418238, + "learning_rate": 1.5991607010944242e-05, + "loss": 0.2607, + "step": 4505 + }, + { + "epoch": 1.9694323144104804, + "grad_norm": 0.23773093829535516, + "learning_rate": 1.5932390049274227e-05, + "loss": 0.2644, + "step": 4510 + }, + { + "epoch": 1.9716157205240175, + "grad_norm": 0.2532308876036313, + "learning_rate": 1.587323161158484e-05, + "loss": 0.272, + "step": 4515 + }, + { + "epoch": 1.9737991266375547, + "grad_norm": 0.2617214304004118, + "learning_rate": 1.5814132079695232e-05, + "loss": 0.2649, + "step": 4520 + }, + { + "epoch": 1.9759825327510918, + "grad_norm": 0.27129287136393265, + "learning_rate": 1.57550918350444e-05, + "loss": 0.2591, + "step": 4525 + }, + { + "epoch": 1.9781659388646289, + "grad_norm": 0.24468851478815726, + "learning_rate": 1.5696111258688657e-05, + "loss": 0.256, + "step": 4530 + }, + { + "epoch": 1.980349344978166, + "grad_norm": 0.23891980205993738, + "learning_rate": 1.563719073129923e-05, + "loss": 0.2655, + "step": 4535 + }, + { + "epoch": 1.982532751091703, + "grad_norm": 0.21564016827597185, + "learning_rate": 1.557833063315976e-05, + "loss": 0.2593, + "step": 4540 + }, + { + "epoch": 1.98471615720524, + "grad_norm": 0.2744337627938704, + "learning_rate": 1.5519531344163873e-05, + "loss": 0.2593, + "step": 4545 + }, + { + "epoch": 1.9868995633187772, + "grad_norm": 0.239539464353039, + "learning_rate": 1.5460793243812732e-05, + "loss": 0.2645, + "step": 4550 + }, + { + "epoch": 1.9890829694323144, + "grad_norm": 0.2652718460025595, + "learning_rate": 1.5402116711212566e-05, + "loss": 0.2712, + "step": 4555 + }, + { + "epoch": 1.9912663755458515, + "grad_norm": 0.23601629641684965, + "learning_rate": 1.5343502125072236e-05, + "loss": 0.2542, + "step": 4560 + }, + { + "epoch": 1.9934497816593888, + "grad_norm": 0.23315305328846694, + "learning_rate": 1.5284949863700805e-05, + "loss": 0.2663, + "step": 4565 + }, + { + "epoch": 1.9956331877729259, + "grad_norm": 0.23782090171706022, + "learning_rate": 1.5226460305005044e-05, + "loss": 0.2622, + "step": 4570 + }, + { + "epoch": 1.997816593886463, + "grad_norm": 0.22388379965681177, + "learning_rate": 1.5168033826487072e-05, + "loss": 0.2692, + "step": 4575 + }, + { + "epoch": 2.0, + "grad_norm": 0.27446739099024114, + "learning_rate": 1.5109670805241846e-05, + "loss": 0.2572, + "step": 4580 + }, + { + "epoch": 2.002183406113537, + "grad_norm": 0.29167629982584226, + "learning_rate": 1.505137161795478e-05, + "loss": 0.2211, + "step": 4585 + }, + { + "epoch": 2.004366812227074, + "grad_norm": 0.2877855784905206, + "learning_rate": 1.4993136640899286e-05, + "loss": 0.2147, + "step": 4590 + }, + { + "epoch": 2.006550218340611, + "grad_norm": 0.2589731295275126, + "learning_rate": 1.4934966249934334e-05, + "loss": 0.2052, + "step": 4595 + }, + { + "epoch": 2.0087336244541483, + "grad_norm": 0.24025382992800187, + "learning_rate": 1.487686082050207e-05, + "loss": 0.2162, + "step": 4600 + }, + { + "epoch": 2.010917030567686, + "grad_norm": 0.22612669576702268, + "learning_rate": 1.4818820727625348e-05, + "loss": 0.2071, + "step": 4605 + }, + { + "epoch": 2.013100436681223, + "grad_norm": 0.2212365851012379, + "learning_rate": 1.4760846345905344e-05, + "loss": 0.2139, + "step": 4610 + }, + { + "epoch": 2.01528384279476, + "grad_norm": 0.20584034227093886, + "learning_rate": 1.4702938049519105e-05, + "loss": 0.2168, + "step": 4615 + }, + { + "epoch": 2.017467248908297, + "grad_norm": 0.21936675852848891, + "learning_rate": 1.4645096212217167e-05, + "loss": 0.2073, + "step": 4620 + }, + { + "epoch": 2.019650655021834, + "grad_norm": 0.26100976524835695, + "learning_rate": 1.4587321207321115e-05, + "loss": 0.2233, + "step": 4625 + }, + { + "epoch": 2.021834061135371, + "grad_norm": 0.2056350483707369, + "learning_rate": 1.4529613407721192e-05, + "loss": 0.2016, + "step": 4630 + }, + { + "epoch": 2.024017467248908, + "grad_norm": 0.21939396929192034, + "learning_rate": 1.4471973185873888e-05, + "loss": 0.2092, + "step": 4635 + }, + { + "epoch": 2.0262008733624453, + "grad_norm": 0.20467895973459885, + "learning_rate": 1.4414400913799535e-05, + "loss": 0.2026, + "step": 4640 + }, + { + "epoch": 2.0283842794759823, + "grad_norm": 0.20303046431819063, + "learning_rate": 1.4356896963079885e-05, + "loss": 0.2034, + "step": 4645 + }, + { + "epoch": 2.03056768558952, + "grad_norm": 0.2147257371596202, + "learning_rate": 1.4299461704855759e-05, + "loss": 0.2038, + "step": 4650 + }, + { + "epoch": 2.032751091703057, + "grad_norm": 0.2709689872203016, + "learning_rate": 1.424209550982461e-05, + "loss": 0.2151, + "step": 4655 + }, + { + "epoch": 2.034934497816594, + "grad_norm": 0.22708728340724726, + "learning_rate": 1.4184798748238148e-05, + "loss": 0.2176, + "step": 4660 + }, + { + "epoch": 2.037117903930131, + "grad_norm": 0.25203147090946515, + "learning_rate": 1.4127571789899946e-05, + "loss": 0.2077, + "step": 4665 + }, + { + "epoch": 2.039301310043668, + "grad_norm": 0.23983946431243502, + "learning_rate": 1.4070415004163056e-05, + "loss": 0.2163, + "step": 4670 + }, + { + "epoch": 2.041484716157205, + "grad_norm": 0.24658657899765385, + "learning_rate": 1.4013328759927625e-05, + "loss": 0.2139, + "step": 4675 + }, + { + "epoch": 2.0436681222707422, + "grad_norm": 0.22543241349761747, + "learning_rate": 1.39563134256385e-05, + "loss": 0.2019, + "step": 4680 + }, + { + "epoch": 2.0458515283842793, + "grad_norm": 0.22236910097875634, + "learning_rate": 1.3899369369282888e-05, + "loss": 0.2183, + "step": 4685 + }, + { + "epoch": 2.0480349344978164, + "grad_norm": 0.20505749283945437, + "learning_rate": 1.3842496958387938e-05, + "loss": 0.2188, + "step": 4690 + }, + { + "epoch": 2.050218340611354, + "grad_norm": 0.2025603121212584, + "learning_rate": 1.3785696560018374e-05, + "loss": 0.2038, + "step": 4695 + }, + { + "epoch": 2.052401746724891, + "grad_norm": 0.2263051334719587, + "learning_rate": 1.372896854077416e-05, + "loss": 0.206, + "step": 4700 + }, + { + "epoch": 2.054585152838428, + "grad_norm": 0.2321594137504455, + "learning_rate": 1.3672313266788106e-05, + "loss": 0.2133, + "step": 4705 + }, + { + "epoch": 2.056768558951965, + "grad_norm": 0.2122041960683085, + "learning_rate": 1.3615731103723507e-05, + "loss": 0.2126, + "step": 4710 + }, + { + "epoch": 2.058951965065502, + "grad_norm": 0.24859114020141224, + "learning_rate": 1.3559222416771793e-05, + "loss": 0.2061, + "step": 4715 + }, + { + "epoch": 2.061135371179039, + "grad_norm": 0.21853834679540102, + "learning_rate": 1.350278757065016e-05, + "loss": 0.2109, + "step": 4720 + }, + { + "epoch": 2.0633187772925763, + "grad_norm": 0.2074388039569142, + "learning_rate": 1.3446426929599229e-05, + "loss": 0.2137, + "step": 4725 + }, + { + "epoch": 2.0655021834061134, + "grad_norm": 0.21863003094886263, + "learning_rate": 1.3390140857380678e-05, + "loss": 0.1967, + "step": 4730 + }, + { + "epoch": 2.067685589519651, + "grad_norm": 0.25260685298764257, + "learning_rate": 1.3333929717274919e-05, + "loss": 0.2255, + "step": 4735 + }, + { + "epoch": 2.069868995633188, + "grad_norm": 0.23960033740421333, + "learning_rate": 1.3277793872078732e-05, + "loss": 0.2195, + "step": 4740 + }, + { + "epoch": 2.072052401746725, + "grad_norm": 0.23761525700313252, + "learning_rate": 1.3221733684102905e-05, + "loss": 0.2128, + "step": 4745 + }, + { + "epoch": 2.074235807860262, + "grad_norm": 0.21987521365657883, + "learning_rate": 1.3165749515169967e-05, + "loss": 0.2115, + "step": 4750 + }, + { + "epoch": 2.076419213973799, + "grad_norm": 0.223938388935573, + "learning_rate": 1.310984172661177e-05, + "loss": 0.2181, + "step": 4755 + }, + { + "epoch": 2.078602620087336, + "grad_norm": 0.2521445985500392, + "learning_rate": 1.305401067926722e-05, + "loss": 0.2143, + "step": 4760 + }, + { + "epoch": 2.0807860262008733, + "grad_norm": 0.23287630698929376, + "learning_rate": 1.2998256733479897e-05, + "loss": 0.2002, + "step": 4765 + }, + { + "epoch": 2.0829694323144103, + "grad_norm": 0.2600254334217533, + "learning_rate": 1.294258024909577e-05, + "loss": 0.2041, + "step": 4770 + }, + { + "epoch": 2.0851528384279474, + "grad_norm": 0.2926480056967554, + "learning_rate": 1.288698158546085e-05, + "loss": 0.2206, + "step": 4775 + }, + { + "epoch": 2.087336244541485, + "grad_norm": 0.21634943547113875, + "learning_rate": 1.2831461101418884e-05, + "loss": 0.2166, + "step": 4780 + }, + { + "epoch": 2.089519650655022, + "grad_norm": 0.2398092797244653, + "learning_rate": 1.2776019155309035e-05, + "loss": 0.2228, + "step": 4785 + }, + { + "epoch": 2.091703056768559, + "grad_norm": 0.22191669799449495, + "learning_rate": 1.2720656104963569e-05, + "loss": 0.2146, + "step": 4790 + }, + { + "epoch": 2.093886462882096, + "grad_norm": 0.22249356654838032, + "learning_rate": 1.2665372307705525e-05, + "loss": 0.2173, + "step": 4795 + }, + { + "epoch": 2.096069868995633, + "grad_norm": 0.2338757269866602, + "learning_rate": 1.2610168120346455e-05, + "loss": 0.2203, + "step": 4800 + }, + { + "epoch": 2.0982532751091703, + "grad_norm": 0.20569835239307332, + "learning_rate": 1.2555043899184077e-05, + "loss": 0.2169, + "step": 4805 + }, + { + "epoch": 2.1004366812227073, + "grad_norm": 0.18840672821070764, + "learning_rate": 1.2500000000000006e-05, + "loss": 0.2131, + "step": 4810 + }, + { + "epoch": 2.1026200873362444, + "grad_norm": 0.229982583270256, + "learning_rate": 1.2445036778057434e-05, + "loss": 0.2155, + "step": 4815 + }, + { + "epoch": 2.1048034934497815, + "grad_norm": 0.21678928738468423, + "learning_rate": 1.2390154588098854e-05, + "loss": 0.2151, + "step": 4820 + }, + { + "epoch": 2.106986899563319, + "grad_norm": 0.20815591769350697, + "learning_rate": 1.233535378434376e-05, + "loss": 0.2075, + "step": 4825 + }, + { + "epoch": 2.109170305676856, + "grad_norm": 0.22153362078682773, + "learning_rate": 1.2280634720486364e-05, + "loss": 0.2136, + "step": 4830 + }, + { + "epoch": 2.111353711790393, + "grad_norm": 0.21312931677811298, + "learning_rate": 1.2225997749693318e-05, + "loss": 0.21, + "step": 4835 + }, + { + "epoch": 2.11353711790393, + "grad_norm": 0.23940821453611505, + "learning_rate": 1.2171443224601443e-05, + "loss": 0.2141, + "step": 4840 + }, + { + "epoch": 2.1157205240174672, + "grad_norm": 0.2242248154195571, + "learning_rate": 1.2116971497315402e-05, + "loss": 0.208, + "step": 4845 + }, + { + "epoch": 2.1179039301310043, + "grad_norm": 0.21221180992392646, + "learning_rate": 1.2062582919405502e-05, + "loss": 0.2093, + "step": 4850 + }, + { + "epoch": 2.1200873362445414, + "grad_norm": 0.22445864486046593, + "learning_rate": 1.200827784190537e-05, + "loss": 0.2146, + "step": 4855 + }, + { + "epoch": 2.1222707423580784, + "grad_norm": 0.2301722383710042, + "learning_rate": 1.1954056615309728e-05, + "loss": 0.2084, + "step": 4860 + }, + { + "epoch": 2.1244541484716155, + "grad_norm": 0.2135558798950853, + "learning_rate": 1.1899919589572092e-05, + "loss": 0.2126, + "step": 4865 + }, + { + "epoch": 2.126637554585153, + "grad_norm": 0.2073308539610393, + "learning_rate": 1.1845867114102532e-05, + "loss": 0.2063, + "step": 4870 + }, + { + "epoch": 2.12882096069869, + "grad_norm": 0.2049312257048255, + "learning_rate": 1.1791899537765416e-05, + "loss": 0.2224, + "step": 4875 + }, + { + "epoch": 2.131004366812227, + "grad_norm": 0.24321581226405323, + "learning_rate": 1.173801720887717e-05, + "loss": 0.2218, + "step": 4880 + }, + { + "epoch": 2.1331877729257642, + "grad_norm": 0.21578319625044598, + "learning_rate": 1.1684220475203986e-05, + "loss": 0.199, + "step": 4885 + }, + { + "epoch": 2.1353711790393013, + "grad_norm": 0.2162898287189938, + "learning_rate": 1.1630509683959643e-05, + "loss": 0.2121, + "step": 4890 + }, + { + "epoch": 2.1375545851528384, + "grad_norm": 0.24512267530291906, + "learning_rate": 1.1576885181803215e-05, + "loss": 0.2171, + "step": 4895 + }, + { + "epoch": 2.1397379912663754, + "grad_norm": 0.19666073074843837, + "learning_rate": 1.1523347314836858e-05, + "loss": 0.2135, + "step": 4900 + }, + { + "epoch": 2.1419213973799125, + "grad_norm": 0.21309169492333768, + "learning_rate": 1.1469896428603563e-05, + "loss": 0.22, + "step": 4905 + }, + { + "epoch": 2.14410480349345, + "grad_norm": 0.2097865812131667, + "learning_rate": 1.1416532868084937e-05, + "loss": 0.2022, + "step": 4910 + }, + { + "epoch": 2.146288209606987, + "grad_norm": 0.22140108936049668, + "learning_rate": 1.1363256977698968e-05, + "loss": 0.2157, + "step": 4915 + }, + { + "epoch": 2.148471615720524, + "grad_norm": 0.20612571861461668, + "learning_rate": 1.1310069101297807e-05, + "loss": 0.212, + "step": 4920 + }, + { + "epoch": 2.150655021834061, + "grad_norm": 0.2172078322760656, + "learning_rate": 1.1256969582165544e-05, + "loss": 0.21, + "step": 4925 + }, + { + "epoch": 2.1528384279475983, + "grad_norm": 0.22655837718087724, + "learning_rate": 1.1203958763016007e-05, + "loss": 0.1969, + "step": 4930 + }, + { + "epoch": 2.1550218340611353, + "grad_norm": 0.23562847936093284, + "learning_rate": 1.1151036985990509e-05, + "loss": 0.209, + "step": 4935 + }, + { + "epoch": 2.1572052401746724, + "grad_norm": 0.21438898804486692, + "learning_rate": 1.1098204592655703e-05, + "loss": 0.2144, + "step": 4940 + }, + { + "epoch": 2.1593886462882095, + "grad_norm": 0.22341501985465956, + "learning_rate": 1.1045461924001325e-05, + "loss": 0.2095, + "step": 4945 + }, + { + "epoch": 2.1615720524017465, + "grad_norm": 0.2154153191432171, + "learning_rate": 1.0992809320438016e-05, + "loss": 0.2071, + "step": 4950 + }, + { + "epoch": 2.1637554585152836, + "grad_norm": 0.19531686390702627, + "learning_rate": 1.0940247121795123e-05, + "loss": 0.2095, + "step": 4955 + }, + { + "epoch": 2.165938864628821, + "grad_norm": 0.22132766807603957, + "learning_rate": 1.08877756673185e-05, + "loss": 0.2174, + "step": 4960 + }, + { + "epoch": 2.168122270742358, + "grad_norm": 0.22517823657634206, + "learning_rate": 1.0835395295668327e-05, + "loss": 0.2179, + "step": 4965 + }, + { + "epoch": 2.1703056768558953, + "grad_norm": 0.21142342412266563, + "learning_rate": 1.0783106344916913e-05, + "loss": 0.2051, + "step": 4970 + }, + { + "epoch": 2.1724890829694323, + "grad_norm": 0.20941158976258883, + "learning_rate": 1.073090915254652e-05, + "loss": 0.2221, + "step": 4975 + }, + { + "epoch": 2.1746724890829694, + "grad_norm": 0.21006000563282354, + "learning_rate": 1.06788040554472e-05, + "loss": 0.2003, + "step": 4980 + }, + { + "epoch": 2.1768558951965065, + "grad_norm": 0.19371851293041617, + "learning_rate": 1.0626791389914573e-05, + "loss": 0.2029, + "step": 4985 + }, + { + "epoch": 2.1790393013100435, + "grad_norm": 0.2202294387310257, + "learning_rate": 1.0574871491647723e-05, + "loss": 0.2141, + "step": 4990 + }, + { + "epoch": 2.1812227074235806, + "grad_norm": 0.22059840659575228, + "learning_rate": 1.0523044695746975e-05, + "loss": 0.2121, + "step": 4995 + }, + { + "epoch": 2.183406113537118, + "grad_norm": 0.2066703461639902, + "learning_rate": 1.0471311336711772e-05, + "loss": 0.2123, + "step": 5000 + }, + { + "epoch": 2.185589519650655, + "grad_norm": 0.20919147411357605, + "learning_rate": 1.0419671748438487e-05, + "loss": 0.2166, + "step": 5005 + }, + { + "epoch": 2.1877729257641922, + "grad_norm": 0.1934804236700962, + "learning_rate": 1.0368126264218282e-05, + "loss": 0.2109, + "step": 5010 + }, + { + "epoch": 2.1899563318777293, + "grad_norm": 0.2089470879409506, + "learning_rate": 1.0316675216734952e-05, + "loss": 0.2057, + "step": 5015 + }, + { + "epoch": 2.1921397379912664, + "grad_norm": 0.19986998684970525, + "learning_rate": 1.026531893806279e-05, + "loss": 0.2225, + "step": 5020 + }, + { + "epoch": 2.1943231441048034, + "grad_norm": 0.22134729099448455, + "learning_rate": 1.0214057759664424e-05, + "loss": 0.2161, + "step": 5025 + }, + { + "epoch": 2.1965065502183405, + "grad_norm": 0.20179325283452076, + "learning_rate": 1.0162892012388703e-05, + "loss": 0.223, + "step": 5030 + }, + { + "epoch": 2.1986899563318776, + "grad_norm": 0.20653676340802546, + "learning_rate": 1.0111822026468515e-05, + "loss": 0.2127, + "step": 5035 + }, + { + "epoch": 2.2008733624454146, + "grad_norm": 0.20118347192898964, + "learning_rate": 1.0060848131518716e-05, + "loss": 0.2132, + "step": 5040 + }, + { + "epoch": 2.203056768558952, + "grad_norm": 0.2127487431663625, + "learning_rate": 1.0009970656533965e-05, + "loss": 0.2138, + "step": 5045 + }, + { + "epoch": 2.2052401746724892, + "grad_norm": 0.18873587826004806, + "learning_rate": 9.95918992988661e-06, + "loss": 0.2159, + "step": 5050 + }, + { + "epoch": 2.2074235807860263, + "grad_norm": 0.21267677980729546, + "learning_rate": 9.908506279324564e-06, + "loss": 0.2094, + "step": 5055 + }, + { + "epoch": 2.2096069868995634, + "grad_norm": 0.20619431733016733, + "learning_rate": 9.857920031969198e-06, + "loss": 0.2066, + "step": 5060 + }, + { + "epoch": 2.2117903930131004, + "grad_norm": 0.2223339638565872, + "learning_rate": 9.807431514313225e-06, + "loss": 0.2315, + "step": 5065 + }, + { + "epoch": 2.2139737991266375, + "grad_norm": 0.2351717184649456, + "learning_rate": 9.757041052218587e-06, + "loss": 0.2192, + "step": 5070 + }, + { + "epoch": 2.2161572052401746, + "grad_norm": 0.18769468217068455, + "learning_rate": 9.70674897091437e-06, + "loss": 0.21, + "step": 5075 + }, + { + "epoch": 2.2183406113537116, + "grad_norm": 0.24549048829311135, + "learning_rate": 9.656555594994687e-06, + "loss": 0.2025, + "step": 5080 + }, + { + "epoch": 2.2205240174672487, + "grad_norm": 0.20415216492897895, + "learning_rate": 9.606461248416573e-06, + "loss": 0.2016, + "step": 5085 + }, + { + "epoch": 2.222707423580786, + "grad_norm": 0.2005430180173854, + "learning_rate": 9.556466254497931e-06, + "loss": 0.2098, + "step": 5090 + }, + { + "epoch": 2.2248908296943233, + "grad_norm": 0.22132570053896475, + "learning_rate": 9.506570935915419e-06, + "loss": 0.2087, + "step": 5095 + }, + { + "epoch": 2.2270742358078603, + "grad_norm": 0.19989818333839046, + "learning_rate": 9.456775614702364e-06, + "loss": 0.2079, + "step": 5100 + }, + { + "epoch": 2.2292576419213974, + "grad_norm": 0.2473690640544336, + "learning_rate": 9.407080612246697e-06, + "loss": 0.2107, + "step": 5105 + }, + { + "epoch": 2.2314410480349345, + "grad_norm": 0.21359326359845074, + "learning_rate": 9.357486249288891e-06, + "loss": 0.2254, + "step": 5110 + }, + { + "epoch": 2.2336244541484715, + "grad_norm": 0.2355252231386045, + "learning_rate": 9.307992845919844e-06, + "loss": 0.2032, + "step": 5115 + }, + { + "epoch": 2.2358078602620086, + "grad_norm": 0.1990769526625258, + "learning_rate": 9.258600721578872e-06, + "loss": 0.2076, + "step": 5120 + }, + { + "epoch": 2.2379912663755457, + "grad_norm": 0.2244729672302675, + "learning_rate": 9.209310195051582e-06, + "loss": 0.2186, + "step": 5125 + }, + { + "epoch": 2.2401746724890828, + "grad_norm": 0.19571847441525664, + "learning_rate": 9.160121584467878e-06, + "loss": 0.1942, + "step": 5130 + }, + { + "epoch": 2.2423580786026203, + "grad_norm": 0.2174940143142323, + "learning_rate": 9.111035207299878e-06, + "loss": 0.2155, + "step": 5135 + }, + { + "epoch": 2.2445414847161573, + "grad_norm": 0.20301405504923226, + "learning_rate": 9.062051380359857e-06, + "loss": 0.2155, + "step": 5140 + }, + { + "epoch": 2.2467248908296944, + "grad_norm": 0.2381569949311249, + "learning_rate": 9.01317041979822e-06, + "loss": 0.2191, + "step": 5145 + }, + { + "epoch": 2.2489082969432315, + "grad_norm": 0.20675760421426403, + "learning_rate": 8.964392641101455e-06, + "loss": 0.198, + "step": 5150 + }, + { + "epoch": 2.2510917030567685, + "grad_norm": 0.18783552296846562, + "learning_rate": 8.915718359090086e-06, + "loss": 0.2049, + "step": 5155 + }, + { + "epoch": 2.2532751091703056, + "grad_norm": 0.20565919831917942, + "learning_rate": 8.867147887916663e-06, + "loss": 0.2227, + "step": 5160 + }, + { + "epoch": 2.2554585152838427, + "grad_norm": 0.205583376077084, + "learning_rate": 8.818681541063714e-06, + "loss": 0.2118, + "step": 5165 + }, + { + "epoch": 2.2576419213973797, + "grad_norm": 0.18747729101469687, + "learning_rate": 8.770319631341745e-06, + "loss": 0.2014, + "step": 5170 + }, + { + "epoch": 2.2598253275109172, + "grad_norm": 0.1903688682819849, + "learning_rate": 8.722062470887169e-06, + "loss": 0.1973, + "step": 5175 + }, + { + "epoch": 2.2620087336244543, + "grad_norm": 0.20544378450917616, + "learning_rate": 8.673910371160368e-06, + "loss": 0.2132, + "step": 5180 + }, + { + "epoch": 2.2641921397379914, + "grad_norm": 0.22300030961664347, + "learning_rate": 8.625863642943623e-06, + "loss": 0.2102, + "step": 5185 + }, + { + "epoch": 2.2663755458515285, + "grad_norm": 0.24719313551585811, + "learning_rate": 8.577922596339142e-06, + "loss": 0.2091, + "step": 5190 + }, + { + "epoch": 2.2685589519650655, + "grad_norm": 0.19395953090113746, + "learning_rate": 8.530087540767029e-06, + "loss": 0.2119, + "step": 5195 + }, + { + "epoch": 2.2707423580786026, + "grad_norm": 0.2057558795183291, + "learning_rate": 8.482358784963312e-06, + "loss": 0.2088, + "step": 5200 + }, + { + "epoch": 2.2729257641921397, + "grad_norm": 0.194298675532194, + "learning_rate": 8.43473663697794e-06, + "loss": 0.1949, + "step": 5205 + }, + { + "epoch": 2.2751091703056767, + "grad_norm": 0.2059192410189242, + "learning_rate": 8.387221404172796e-06, + "loss": 0.2036, + "step": 5210 + }, + { + "epoch": 2.277292576419214, + "grad_norm": 0.20744243911102733, + "learning_rate": 8.339813393219714e-06, + "loss": 0.2146, + "step": 5215 + }, + { + "epoch": 2.279475982532751, + "grad_norm": 0.2043265849228992, + "learning_rate": 8.292512910098508e-06, + "loss": 0.2206, + "step": 5220 + }, + { + "epoch": 2.2816593886462884, + "grad_norm": 0.2100815969871642, + "learning_rate": 8.245320260094955e-06, + "loss": 0.2199, + "step": 5225 + }, + { + "epoch": 2.2838427947598254, + "grad_norm": 0.20976956534920874, + "learning_rate": 8.198235747798893e-06, + "loss": 0.2168, + "step": 5230 + }, + { + "epoch": 2.2860262008733625, + "grad_norm": 0.2050232275697135, + "learning_rate": 8.151259677102211e-06, + "loss": 0.216, + "step": 5235 + }, + { + "epoch": 2.2882096069868996, + "grad_norm": 4.122598477737714, + "learning_rate": 8.104392351196885e-06, + "loss": 0.2198, + "step": 5240 + }, + { + "epoch": 2.2903930131004366, + "grad_norm": 0.2237909328728828, + "learning_rate": 8.057634072573048e-06, + "loss": 0.2173, + "step": 5245 + }, + { + "epoch": 2.2925764192139737, + "grad_norm": 0.21176821690953787, + "learning_rate": 8.010985143017008e-06, + "loss": 0.2157, + "step": 5250 + }, + { + "epoch": 2.2947598253275108, + "grad_norm": 0.8545824488764707, + "learning_rate": 7.964445863609329e-06, + "loss": 0.215, + "step": 5255 + }, + { + "epoch": 2.2969432314410483, + "grad_norm": 0.23797387294881264, + "learning_rate": 7.91801653472286e-06, + "loss": 0.1993, + "step": 5260 + }, + { + "epoch": 2.2991266375545854, + "grad_norm": 0.209126744021443, + "learning_rate": 7.871697456020819e-06, + "loss": 0.2285, + "step": 5265 + }, + { + "epoch": 2.3013100436681224, + "grad_norm": 0.2036921534501079, + "learning_rate": 7.825488926454851e-06, + "loss": 0.2137, + "step": 5270 + }, + { + "epoch": 2.3034934497816595, + "grad_norm": 0.19891864557452632, + "learning_rate": 7.77939124426308e-06, + "loss": 0.2094, + "step": 5275 + }, + { + "epoch": 2.3056768558951966, + "grad_norm": 0.21772742093397335, + "learning_rate": 7.733404706968215e-06, + "loss": 0.2097, + "step": 5280 + }, + { + "epoch": 2.3078602620087336, + "grad_norm": 0.1920838796289188, + "learning_rate": 7.687529611375616e-06, + "loss": 0.2161, + "step": 5285 + }, + { + "epoch": 2.3100436681222707, + "grad_norm": 0.20114220947132505, + "learning_rate": 7.64176625357138e-06, + "loss": 0.2041, + "step": 5290 + }, + { + "epoch": 2.3122270742358078, + "grad_norm": 0.20058172964283594, + "learning_rate": 7.596114928920422e-06, + "loss": 0.2124, + "step": 5295 + }, + { + "epoch": 2.314410480349345, + "grad_norm": 0.20110764788896288, + "learning_rate": 7.55057593206458e-06, + "loss": 0.2018, + "step": 5300 + }, + { + "epoch": 2.316593886462882, + "grad_norm": 0.2033865906846183, + "learning_rate": 7.505149556920699e-06, + "loss": 0.2139, + "step": 5305 + }, + { + "epoch": 2.3187772925764194, + "grad_norm": 0.21910248265310586, + "learning_rate": 7.459836096678757e-06, + "loss": 0.2153, + "step": 5310 + }, + { + "epoch": 2.3209606986899565, + "grad_norm": 0.2026773384631961, + "learning_rate": 7.41463584379995e-06, + "loss": 0.215, + "step": 5315 + }, + { + "epoch": 2.3231441048034935, + "grad_norm": 0.2127225287015194, + "learning_rate": 7.369549090014821e-06, + "loss": 0.212, + "step": 5320 + }, + { + "epoch": 2.3253275109170306, + "grad_norm": 0.20869494683552092, + "learning_rate": 7.32457612632135e-06, + "loss": 0.216, + "step": 5325 + }, + { + "epoch": 2.3275109170305677, + "grad_norm": 0.19166630407586727, + "learning_rate": 7.279717242983117e-06, + "loss": 0.2166, + "step": 5330 + }, + { + "epoch": 2.3296943231441047, + "grad_norm": 0.2004851412037729, + "learning_rate": 7.234972729527401e-06, + "loss": 0.207, + "step": 5335 + }, + { + "epoch": 2.331877729257642, + "grad_norm": 0.18554259866541592, + "learning_rate": 7.190342874743311e-06, + "loss": 0.201, + "step": 5340 + }, + { + "epoch": 2.334061135371179, + "grad_norm": 0.21209107877243255, + "learning_rate": 7.145827966679936e-06, + "loss": 0.1968, + "step": 5345 + }, + { + "epoch": 2.3362445414847164, + "grad_norm": 0.20153718074405388, + "learning_rate": 7.101428292644477e-06, + "loss": 0.2176, + "step": 5350 + }, + { + "epoch": 2.3384279475982535, + "grad_norm": 0.18992678255749726, + "learning_rate": 7.057144139200378e-06, + "loss": 0.2013, + "step": 5355 + }, + { + "epoch": 2.3406113537117905, + "grad_norm": 0.1830289392115265, + "learning_rate": 7.0129757921655355e-06, + "loss": 0.1979, + "step": 5360 + }, + { + "epoch": 2.3427947598253276, + "grad_norm": 0.19740674215886508, + "learning_rate": 6.968923536610356e-06, + "loss": 0.2078, + "step": 5365 + }, + { + "epoch": 2.3449781659388647, + "grad_norm": 0.2121990875989476, + "learning_rate": 6.924987656856002e-06, + "loss": 0.1936, + "step": 5370 + }, + { + "epoch": 2.3471615720524017, + "grad_norm": 0.22075994643117536, + "learning_rate": 6.881168436472524e-06, + "loss": 0.2114, + "step": 5375 + }, + { + "epoch": 2.349344978165939, + "grad_norm": 0.20805940844514723, + "learning_rate": 6.837466158277025e-06, + "loss": 0.1941, + "step": 5380 + }, + { + "epoch": 2.351528384279476, + "grad_norm": 0.18880443913500422, + "learning_rate": 6.7938811043318475e-06, + "loss": 0.2151, + "step": 5385 + }, + { + "epoch": 2.353711790393013, + "grad_norm": 0.23050444783405694, + "learning_rate": 6.750413555942742e-06, + "loss": 0.2066, + "step": 5390 + }, + { + "epoch": 2.35589519650655, + "grad_norm": 0.19659728337288793, + "learning_rate": 6.707063793657065e-06, + "loss": 0.2168, + "step": 5395 + }, + { + "epoch": 2.3580786026200875, + "grad_norm": 0.2044183886617955, + "learning_rate": 6.663832097261952e-06, + "loss": 0.2193, + "step": 5400 + }, + { + "epoch": 2.3602620087336246, + "grad_norm": 0.19473776562391912, + "learning_rate": 6.620718745782525e-06, + "loss": 0.2031, + "step": 5405 + }, + { + "epoch": 2.3624454148471616, + "grad_norm": 0.19433516742290474, + "learning_rate": 6.577724017480094e-06, + "loss": 0.2147, + "step": 5410 + }, + { + "epoch": 2.3646288209606987, + "grad_norm": 0.2002897275632141, + "learning_rate": 6.534848189850329e-06, + "loss": 0.1992, + "step": 5415 + }, + { + "epoch": 2.3668122270742358, + "grad_norm": 0.1981020777647114, + "learning_rate": 6.492091539621517e-06, + "loss": 0.2029, + "step": 5420 + }, + { + "epoch": 2.368995633187773, + "grad_norm": 0.18263004359615023, + "learning_rate": 6.449454342752748e-06, + "loss": 0.2018, + "step": 5425 + }, + { + "epoch": 2.37117903930131, + "grad_norm": 0.1878131474845262, + "learning_rate": 6.406936874432134e-06, + "loss": 0.2037, + "step": 5430 + }, + { + "epoch": 2.373362445414847, + "grad_norm": 0.204062897390976, + "learning_rate": 6.364539409075043e-06, + "loss": 0.2017, + "step": 5435 + }, + { + "epoch": 2.3755458515283845, + "grad_norm": 0.19374315646095067, + "learning_rate": 6.322262220322314e-06, + "loss": 0.2093, + "step": 5440 + }, + { + "epoch": 2.3777292576419216, + "grad_norm": 0.20250412793089098, + "learning_rate": 6.280105581038509e-06, + "loss": 0.2031, + "step": 5445 + }, + { + "epoch": 2.3799126637554586, + "grad_norm": 0.18639353569932318, + "learning_rate": 6.238069763310136e-06, + "loss": 0.2082, + "step": 5450 + }, + { + "epoch": 2.3820960698689957, + "grad_norm": 0.19128091651488877, + "learning_rate": 6.1961550384438984e-06, + "loss": 0.2066, + "step": 5455 + }, + { + "epoch": 2.3842794759825328, + "grad_norm": 0.19608351533598506, + "learning_rate": 6.154361676964956e-06, + "loss": 0.2165, + "step": 5460 + }, + { + "epoch": 2.38646288209607, + "grad_norm": 0.18656964338831433, + "learning_rate": 6.112689948615144e-06, + "loss": 0.2019, + "step": 5465 + }, + { + "epoch": 2.388646288209607, + "grad_norm": 0.18488662819515192, + "learning_rate": 6.071140122351276e-06, + "loss": 0.2018, + "step": 5470 + }, + { + "epoch": 2.390829694323144, + "grad_norm": 0.19839674782985758, + "learning_rate": 6.029712466343382e-06, + "loss": 0.2057, + "step": 5475 + }, + { + "epoch": 2.393013100436681, + "grad_norm": 0.1935316894500808, + "learning_rate": 5.988407247972982e-06, + "loss": 0.2101, + "step": 5480 + }, + { + "epoch": 2.395196506550218, + "grad_norm": 0.19014181610486888, + "learning_rate": 5.947224733831364e-06, + "loss": 0.214, + "step": 5485 + }, + { + "epoch": 2.3973799126637556, + "grad_norm": 0.2047537532894175, + "learning_rate": 5.906165189717855e-06, + "loss": 0.2069, + "step": 5490 + }, + { + "epoch": 2.3995633187772927, + "grad_norm": 0.20244514937691652, + "learning_rate": 5.8652288806381215e-06, + "loss": 0.2093, + "step": 5495 + }, + { + "epoch": 2.4017467248908297, + "grad_norm": 0.2029682194929424, + "learning_rate": 5.824416070802438e-06, + "loss": 0.1989, + "step": 5500 + }, + { + "epoch": 2.403930131004367, + "grad_norm": 0.19862209705141703, + "learning_rate": 5.7837270236239995e-06, + "loss": 0.1988, + "step": 5505 + }, + { + "epoch": 2.406113537117904, + "grad_norm": 0.19816634806985825, + "learning_rate": 5.74316200171722e-06, + "loss": 0.2042, + "step": 5510 + }, + { + "epoch": 2.408296943231441, + "grad_norm": 0.20129123615306388, + "learning_rate": 5.70272126689601e-06, + "loss": 0.2036, + "step": 5515 + }, + { + "epoch": 2.410480349344978, + "grad_norm": 0.18987017520409014, + "learning_rate": 5.662405080172134e-06, + "loss": 0.2078, + "step": 5520 + }, + { + "epoch": 2.4126637554585155, + "grad_norm": 0.19052375793050977, + "learning_rate": 5.622213701753487e-06, + "loss": 0.2231, + "step": 5525 + }, + { + "epoch": 2.4148471615720526, + "grad_norm": 0.19323234507341908, + "learning_rate": 5.582147391042433e-06, + "loss": 0.2092, + "step": 5530 + }, + { + "epoch": 2.4170305676855897, + "grad_norm": 0.1924708259179898, + "learning_rate": 5.542206406634129e-06, + "loss": 0.2078, + "step": 5535 + }, + { + "epoch": 2.4192139737991267, + "grad_norm": 0.207006675227268, + "learning_rate": 5.5023910063148445e-06, + "loss": 0.2109, + "step": 5540 + }, + { + "epoch": 2.421397379912664, + "grad_norm": 0.2002952114074735, + "learning_rate": 5.46270144706032e-06, + "loss": 0.198, + "step": 5545 + }, + { + "epoch": 2.423580786026201, + "grad_norm": 0.2091856914324174, + "learning_rate": 5.423137985034085e-06, + "loss": 0.2138, + "step": 5550 + }, + { + "epoch": 2.425764192139738, + "grad_norm": 0.22941186423542465, + "learning_rate": 5.3837008755858145e-06, + "loss": 0.2013, + "step": 5555 + }, + { + "epoch": 2.427947598253275, + "grad_norm": 0.1974516927188841, + "learning_rate": 5.3443903732496966e-06, + "loss": 0.2069, + "step": 5560 + }, + { + "epoch": 2.430131004366812, + "grad_norm": 0.1861177274579643, + "learning_rate": 5.305206731742748e-06, + "loss": 0.2086, + "step": 5565 + }, + { + "epoch": 2.432314410480349, + "grad_norm": 0.196399769855382, + "learning_rate": 5.266150203963224e-06, + "loss": 0.2155, + "step": 5570 + }, + { + "epoch": 2.4344978165938866, + "grad_norm": 0.20583463409268135, + "learning_rate": 5.227221041988956e-06, + "loss": 0.2055, + "step": 5575 + }, + { + "epoch": 2.4366812227074237, + "grad_norm": 0.19145217163247738, + "learning_rate": 5.188419497075739e-06, + "loss": 0.2045, + "step": 5580 + }, + { + "epoch": 2.438864628820961, + "grad_norm": 0.2300625133365478, + "learning_rate": 5.1497458196556995e-06, + "loss": 0.2192, + "step": 5585 + }, + { + "epoch": 2.441048034934498, + "grad_norm": 0.19252331533699785, + "learning_rate": 5.111200259335688e-06, + "loss": 0.2089, + "step": 5590 + }, + { + "epoch": 2.443231441048035, + "grad_norm": 0.20014832340012859, + "learning_rate": 5.072783064895664e-06, + "loss": 0.2243, + "step": 5595 + }, + { + "epoch": 2.445414847161572, + "grad_norm": 0.19507458711265246, + "learning_rate": 5.034494484287086e-06, + "loss": 0.2195, + "step": 5600 + }, + { + "epoch": 2.447598253275109, + "grad_norm": 0.18618941255923363, + "learning_rate": 4.996334764631322e-06, + "loss": 0.2034, + "step": 5605 + }, + { + "epoch": 2.449781659388646, + "grad_norm": 0.2223628562088218, + "learning_rate": 4.958304152218046e-06, + "loss": 0.2076, + "step": 5610 + }, + { + "epoch": 2.4519650655021836, + "grad_norm": 0.22749716704810904, + "learning_rate": 4.920402892503648e-06, + "loss": 0.2142, + "step": 5615 + }, + { + "epoch": 2.4541484716157207, + "grad_norm": 0.20326433502448837, + "learning_rate": 4.882631230109655e-06, + "loss": 0.206, + "step": 5620 + }, + { + "epoch": 2.4563318777292578, + "grad_norm": 0.20525401555309347, + "learning_rate": 4.844989408821149e-06, + "loss": 0.2042, + "step": 5625 + }, + { + "epoch": 2.458515283842795, + "grad_norm": 0.20210008215232908, + "learning_rate": 4.807477671585195e-06, + "loss": 0.1999, + "step": 5630 + }, + { + "epoch": 2.460698689956332, + "grad_norm": 0.21710829320874975, + "learning_rate": 4.770096260509274e-06, + "loss": 0.203, + "step": 5635 + }, + { + "epoch": 2.462882096069869, + "grad_norm": 0.18303629687950113, + "learning_rate": 4.732845416859713e-06, + "loss": 0.2134, + "step": 5640 + }, + { + "epoch": 2.465065502183406, + "grad_norm": 0.19284778474305014, + "learning_rate": 4.695725381060137e-06, + "loss": 0.2062, + "step": 5645 + }, + { + "epoch": 2.467248908296943, + "grad_norm": 0.18825083490369524, + "learning_rate": 4.658736392689922e-06, + "loss": 0.2214, + "step": 5650 + }, + { + "epoch": 2.46943231441048, + "grad_norm": 0.20310843234802933, + "learning_rate": 4.6218786904826175e-06, + "loss": 0.2132, + "step": 5655 + }, + { + "epoch": 2.4716157205240172, + "grad_norm": 0.21879459935317394, + "learning_rate": 4.58515251232445e-06, + "loss": 0.2193, + "step": 5660 + }, + { + "epoch": 2.4737991266375547, + "grad_norm": 0.2064948920674535, + "learning_rate": 4.548558095252758e-06, + "loss": 0.206, + "step": 5665 + }, + { + "epoch": 2.475982532751092, + "grad_norm": 0.20819617169964372, + "learning_rate": 4.512095675454472e-06, + "loss": 0.2116, + "step": 5670 + }, + { + "epoch": 2.478165938864629, + "grad_norm": 0.19013108098717665, + "learning_rate": 4.475765488264588e-06, + "loss": 0.2031, + "step": 5675 + }, + { + "epoch": 2.480349344978166, + "grad_norm": 0.19875013946254505, + "learning_rate": 4.439567768164651e-06, + "loss": 0.2017, + "step": 5680 + }, + { + "epoch": 2.482532751091703, + "grad_norm": 0.235432313345855, + "learning_rate": 4.4035027487812325e-06, + "loss": 0.2119, + "step": 5685 + }, + { + "epoch": 2.48471615720524, + "grad_norm": 0.19883691503756268, + "learning_rate": 4.367570662884438e-06, + "loss": 0.1997, + "step": 5690 + }, + { + "epoch": 2.486899563318777, + "grad_norm": 0.19289587740272107, + "learning_rate": 4.331771742386395e-06, + "loss": 0.2096, + "step": 5695 + }, + { + "epoch": 2.489082969432314, + "grad_norm": 0.18568302603914705, + "learning_rate": 4.296106218339757e-06, + "loss": 0.2056, + "step": 5700 + }, + { + "epoch": 2.4912663755458517, + "grad_norm": 0.1886391942932925, + "learning_rate": 4.260574320936203e-06, + "loss": 0.2022, + "step": 5705 + }, + { + "epoch": 2.493449781659389, + "grad_norm": 0.19736454250903648, + "learning_rate": 4.225176279504975e-06, + "loss": 0.2139, + "step": 5710 + }, + { + "epoch": 2.495633187772926, + "grad_norm": 0.20848042828630423, + "learning_rate": 4.1899123225113825e-06, + "loss": 0.2033, + "step": 5715 + }, + { + "epoch": 2.497816593886463, + "grad_norm": 0.18156063895028207, + "learning_rate": 4.154782677555324e-06, + "loss": 0.2055, + "step": 5720 + }, + { + "epoch": 2.5, + "grad_norm": 0.1915665995869679, + "learning_rate": 4.119787571369829e-06, + "loss": 0.2056, + "step": 5725 + }, + { + "epoch": 2.502183406113537, + "grad_norm": 0.20789030655261928, + "learning_rate": 4.084927229819591e-06, + "loss": 0.1969, + "step": 5730 + }, + { + "epoch": 2.504366812227074, + "grad_norm": 0.19567843626163597, + "learning_rate": 4.050201877899501e-06, + "loss": 0.1967, + "step": 5735 + }, + { + "epoch": 2.506550218340611, + "grad_norm": 0.19524615840080548, + "learning_rate": 4.0156117397332074e-06, + "loss": 0.2017, + "step": 5740 + }, + { + "epoch": 2.5087336244541483, + "grad_norm": 0.1851828139613455, + "learning_rate": 3.9811570385716705e-06, + "loss": 0.2187, + "step": 5745 + }, + { + "epoch": 2.5109170305676853, + "grad_norm": 0.18768129563315683, + "learning_rate": 3.9468379967917025e-06, + "loss": 0.2139, + "step": 5750 + }, + { + "epoch": 2.513100436681223, + "grad_norm": 0.1908316174021218, + "learning_rate": 3.912654835894564e-06, + "loss": 0.2085, + "step": 5755 + }, + { + "epoch": 2.51528384279476, + "grad_norm": 0.18887997682497717, + "learning_rate": 3.878607776504489e-06, + "loss": 0.2089, + "step": 5760 + }, + { + "epoch": 2.517467248908297, + "grad_norm": 0.18931507612361673, + "learning_rate": 3.844697038367314e-06, + "loss": 0.2118, + "step": 5765 + }, + { + "epoch": 2.519650655021834, + "grad_norm": 0.1955389549996342, + "learning_rate": 3.8109228403490268e-06, + "loss": 0.2041, + "step": 5770 + }, + { + "epoch": 2.521834061135371, + "grad_norm": 0.1985311672510908, + "learning_rate": 3.7772854004343494e-06, + "loss": 0.2117, + "step": 5775 + }, + { + "epoch": 2.524017467248908, + "grad_norm": 0.18193407723396662, + "learning_rate": 3.7437849357253555e-06, + "loss": 0.2051, + "step": 5780 + }, + { + "epoch": 2.5262008733624453, + "grad_norm": 0.19363438642248365, + "learning_rate": 3.7104216624400502e-06, + "loss": 0.2112, + "step": 5785 + }, + { + "epoch": 2.5283842794759828, + "grad_norm": 0.18728937794992326, + "learning_rate": 3.677195795910984e-06, + "loss": 0.2052, + "step": 5790 + }, + { + "epoch": 2.53056768558952, + "grad_norm": 0.17522592702667172, + "learning_rate": 3.6441075505838586e-06, + "loss": 0.2108, + "step": 5795 + }, + { + "epoch": 2.532751091703057, + "grad_norm": 0.19669895879396115, + "learning_rate": 3.6111571400161387e-06, + "loss": 0.2107, + "step": 5800 + }, + { + "epoch": 2.534934497816594, + "grad_norm": 0.2086966810335374, + "learning_rate": 3.578344776875689e-06, + "loss": 0.2025, + "step": 5805 + }, + { + "epoch": 2.537117903930131, + "grad_norm": 0.18427750014405722, + "learning_rate": 3.5456706729393774e-06, + "loss": 0.2089, + "step": 5810 + }, + { + "epoch": 2.539301310043668, + "grad_norm": 0.19759644279021854, + "learning_rate": 3.5131350390917333e-06, + "loss": 0.2136, + "step": 5815 + }, + { + "epoch": 2.541484716157205, + "grad_norm": 0.18832000886509567, + "learning_rate": 3.4807380853235704e-06, + "loss": 0.2206, + "step": 5820 + }, + { + "epoch": 2.5436681222707422, + "grad_norm": 0.1939256790542837, + "learning_rate": 3.4484800207306405e-06, + "loss": 0.2104, + "step": 5825 + }, + { + "epoch": 2.5458515283842793, + "grad_norm": 0.1981092726688764, + "learning_rate": 3.4163610535122738e-06, + "loss": 0.2082, + "step": 5830 + }, + { + "epoch": 2.5480349344978164, + "grad_norm": 0.20236118691003865, + "learning_rate": 3.384381390970048e-06, + "loss": 0.2046, + "step": 5835 + }, + { + "epoch": 2.5502183406113534, + "grad_norm": 0.19004473471111177, + "learning_rate": 3.3525412395064375e-06, + "loss": 0.2023, + "step": 5840 + }, + { + "epoch": 2.552401746724891, + "grad_norm": 0.20444360368222714, + "learning_rate": 3.32084080462349e-06, + "loss": 0.2027, + "step": 5845 + }, + { + "epoch": 2.554585152838428, + "grad_norm": 0.2175490290108786, + "learning_rate": 3.2892802909214955e-06, + "loss": 0.2096, + "step": 5850 + }, + { + "epoch": 2.556768558951965, + "grad_norm": 0.19247531436865034, + "learning_rate": 3.2578599020976707e-06, + "loss": 0.1951, + "step": 5855 + }, + { + "epoch": 2.558951965065502, + "grad_norm": 0.1874278832882864, + "learning_rate": 3.2265798409448376e-06, + "loss": 0.211, + "step": 5860 + }, + { + "epoch": 2.561135371179039, + "grad_norm": 0.19407368076943168, + "learning_rate": 3.1954403093501227e-06, + "loss": 0.2179, + "step": 5865 + }, + { + "epoch": 2.5633187772925763, + "grad_norm": 0.176804419787979, + "learning_rate": 3.1644415082936417e-06, + "loss": 0.2017, + "step": 5870 + }, + { + "epoch": 2.565502183406114, + "grad_norm": 0.1936425739896731, + "learning_rate": 3.1335836378472233e-06, + "loss": 0.1963, + "step": 5875 + }, + { + "epoch": 2.567685589519651, + "grad_norm": 0.18623478504104157, + "learning_rate": 3.10286689717309e-06, + "loss": 0.2076, + "step": 5880 + }, + { + "epoch": 2.569868995633188, + "grad_norm": 0.18973768721599718, + "learning_rate": 3.0722914845225963e-06, + "loss": 0.211, + "step": 5885 + }, + { + "epoch": 2.572052401746725, + "grad_norm": 0.1887388707833972, + "learning_rate": 3.04185759723494e-06, + "loss": 0.1983, + "step": 5890 + }, + { + "epoch": 2.574235807860262, + "grad_norm": 0.18187528811137918, + "learning_rate": 3.0115654317358884e-06, + "loss": 0.2119, + "step": 5895 + }, + { + "epoch": 2.576419213973799, + "grad_norm": 0.19057590033301217, + "learning_rate": 2.981415183536504e-06, + "loss": 0.212, + "step": 5900 + }, + { + "epoch": 2.578602620087336, + "grad_norm": 0.18728042390500685, + "learning_rate": 2.9514070472318965e-06, + "loss": 0.1956, + "step": 5905 + }, + { + "epoch": 2.5807860262008733, + "grad_norm": 0.18967646290907864, + "learning_rate": 2.9215412164999594e-06, + "loss": 0.2006, + "step": 5910 + }, + { + "epoch": 2.5829694323144103, + "grad_norm": 0.19120785023883197, + "learning_rate": 2.891817884100123e-06, + "loss": 0.192, + "step": 5915 + }, + { + "epoch": 2.5851528384279474, + "grad_norm": 0.18671520762510882, + "learning_rate": 2.862237241872101e-06, + "loss": 0.2215, + "step": 5920 + }, + { + "epoch": 2.5873362445414845, + "grad_norm": 0.1891615762513396, + "learning_rate": 2.832799480734663e-06, + "loss": 0.2244, + "step": 5925 + }, + { + "epoch": 2.589519650655022, + "grad_norm": 0.1807780047483408, + "learning_rate": 2.8035047906844e-06, + "loss": 0.2053, + "step": 5930 + }, + { + "epoch": 2.591703056768559, + "grad_norm": 0.18284608120565451, + "learning_rate": 2.774353360794493e-06, + "loss": 0.2047, + "step": 5935 + }, + { + "epoch": 2.593886462882096, + "grad_norm": 0.18110685745874158, + "learning_rate": 2.7453453792134986e-06, + "loss": 0.213, + "step": 5940 + }, + { + "epoch": 2.596069868995633, + "grad_norm": 0.18936818198939037, + "learning_rate": 2.7164810331641366e-06, + "loss": 0.2038, + "step": 5945 + }, + { + "epoch": 2.5982532751091703, + "grad_norm": 0.19029850620608788, + "learning_rate": 2.687760508942064e-06, + "loss": 0.2116, + "step": 5950 + }, + { + "epoch": 2.6004366812227073, + "grad_norm": 0.19115405588229892, + "learning_rate": 2.659183991914696e-06, + "loss": 0.2145, + "step": 5955 + }, + { + "epoch": 2.6026200873362444, + "grad_norm": 0.1945513238775359, + "learning_rate": 2.6307516665200012e-06, + "loss": 0.208, + "step": 5960 + }, + { + "epoch": 2.604803493449782, + "grad_norm": 0.19282829361579937, + "learning_rate": 2.602463716265302e-06, + "loss": 0.2179, + "step": 5965 + }, + { + "epoch": 2.606986899563319, + "grad_norm": 0.2002570604911309, + "learning_rate": 2.5743203237261065e-06, + "loss": 0.2095, + "step": 5970 + }, + { + "epoch": 2.609170305676856, + "grad_norm": 0.18334332369689957, + "learning_rate": 2.5463216705449155e-06, + "loss": 0.2185, + "step": 5975 + }, + { + "epoch": 2.611353711790393, + "grad_norm": 0.181736048243745, + "learning_rate": 2.5184679374300553e-06, + "loss": 0.2085, + "step": 5980 + }, + { + "epoch": 2.61353711790393, + "grad_norm": 0.17385940038429637, + "learning_rate": 2.4907593041545163e-06, + "loss": 0.2167, + "step": 5985 + }, + { + "epoch": 2.6157205240174672, + "grad_norm": 0.17922911028497665, + "learning_rate": 2.4631959495547833e-06, + "loss": 0.1995, + "step": 5990 + }, + { + "epoch": 2.6179039301310043, + "grad_norm": 0.17779619304763405, + "learning_rate": 2.4357780515297e-06, + "loss": 0.1941, + "step": 5995 + }, + { + "epoch": 2.6200873362445414, + "grad_norm": 0.19092236134949003, + "learning_rate": 2.408505787039281e-06, + "loss": 0.2023, + "step": 6000 + }, + { + "epoch": 2.6222707423580784, + "grad_norm": 0.21537785339457424, + "learning_rate": 2.381379332103623e-06, + "loss": 0.2039, + "step": 6005 + }, + { + "epoch": 2.6244541484716155, + "grad_norm": 0.19076636501936398, + "learning_rate": 2.3543988618017237e-06, + "loss": 0.2007, + "step": 6010 + }, + { + "epoch": 2.6266375545851526, + "grad_norm": 0.18091299115157525, + "learning_rate": 2.327564550270381e-06, + "loss": 0.2003, + "step": 6015 + }, + { + "epoch": 2.62882096069869, + "grad_norm": 0.17652429068296252, + "learning_rate": 2.3008765707030532e-06, + "loss": 0.1987, + "step": 6020 + }, + { + "epoch": 2.631004366812227, + "grad_norm": 0.19337565491737743, + "learning_rate": 2.2743350953487424e-06, + "loss": 0.2161, + "step": 6025 + }, + { + "epoch": 2.6331877729257642, + "grad_norm": 0.19521279815086381, + "learning_rate": 2.2479402955108945e-06, + "loss": 0.1994, + "step": 6030 + }, + { + "epoch": 2.6353711790393013, + "grad_norm": 0.18448803162094932, + "learning_rate": 2.2216923415462802e-06, + "loss": 0.2071, + "step": 6035 + }, + { + "epoch": 2.6375545851528384, + "grad_norm": 0.19952616634096593, + "learning_rate": 2.1955914028639e-06, + "loss": 0.1997, + "step": 6040 + }, + { + "epoch": 2.6397379912663754, + "grad_norm": 0.17820128872734733, + "learning_rate": 2.1696376479238947e-06, + "loss": 0.2095, + "step": 6045 + }, + { + "epoch": 2.641921397379913, + "grad_norm": 0.18942931129920965, + "learning_rate": 2.1438312442364467e-06, + "loss": 0.2065, + "step": 6050 + }, + { + "epoch": 2.64410480349345, + "grad_norm": 0.1880722865584005, + "learning_rate": 2.1181723583607177e-06, + "loss": 0.2011, + "step": 6055 + }, + { + "epoch": 2.646288209606987, + "grad_norm": 0.19525082185206008, + "learning_rate": 2.092661155903758e-06, + "loss": 0.2075, + "step": 6060 + }, + { + "epoch": 2.648471615720524, + "grad_norm": 0.17350084934940418, + "learning_rate": 2.067297801519441e-06, + "loss": 0.1998, + "step": 6065 + }, + { + "epoch": 2.650655021834061, + "grad_norm": 0.18323734039467143, + "learning_rate": 2.0420824589074073e-06, + "loss": 0.2066, + "step": 6070 + }, + { + "epoch": 2.6528384279475983, + "grad_norm": 0.19042032301324274, + "learning_rate": 2.0170152908119964e-06, + "loss": 0.2078, + "step": 6075 + }, + { + "epoch": 2.6550218340611353, + "grad_norm": 0.1822344566119162, + "learning_rate": 1.9920964590212088e-06, + "loss": 0.2026, + "step": 6080 + }, + { + "epoch": 2.6572052401746724, + "grad_norm": 0.18350664419747995, + "learning_rate": 1.967326124365651e-06, + "loss": 0.2046, + "step": 6085 + }, + { + "epoch": 2.6593886462882095, + "grad_norm": 0.17900525181219676, + "learning_rate": 1.9427044467175054e-06, + "loss": 0.2053, + "step": 6090 + }, + { + "epoch": 2.6615720524017465, + "grad_norm": 0.18662259458308622, + "learning_rate": 1.9182315849894996e-06, + "loss": 0.1969, + "step": 6095 + }, + { + "epoch": 2.6637554585152836, + "grad_norm": 0.18210726045913153, + "learning_rate": 1.8939076971338592e-06, + "loss": 0.2133, + "step": 6100 + }, + { + "epoch": 2.665938864628821, + "grad_norm": 0.18936892016307003, + "learning_rate": 1.8697329401413333e-06, + "loss": 0.2036, + "step": 6105 + }, + { + "epoch": 2.668122270742358, + "grad_norm": 0.17873699633744589, + "learning_rate": 1.8457074700401284e-06, + "loss": 0.2015, + "step": 6110 + }, + { + "epoch": 2.6703056768558953, + "grad_norm": 0.19259954300544802, + "learning_rate": 1.821831441894939e-06, + "loss": 0.2168, + "step": 6115 + }, + { + "epoch": 2.6724890829694323, + "grad_norm": 0.1765084388420357, + "learning_rate": 1.7981050098059364e-06, + "loss": 0.2051, + "step": 6120 + }, + { + "epoch": 2.6746724890829694, + "grad_norm": 0.18953853048678723, + "learning_rate": 1.7745283269077622e-06, + "loss": 0.1945, + "step": 6125 + }, + { + "epoch": 2.6768558951965065, + "grad_norm": 0.17824046665997959, + "learning_rate": 1.751101545368558e-06, + "loss": 0.1987, + "step": 6130 + }, + { + "epoch": 2.6790393013100435, + "grad_norm": 0.1839474436743472, + "learning_rate": 1.7278248163889787e-06, + "loss": 0.2297, + "step": 6135 + }, + { + "epoch": 2.681222707423581, + "grad_norm": 0.198016592733994, + "learning_rate": 1.7046982902011954e-06, + "loss": 0.2004, + "step": 6140 + }, + { + "epoch": 2.683406113537118, + "grad_norm": 0.1754478072579575, + "learning_rate": 1.6817221160679603e-06, + "loss": 0.2165, + "step": 6145 + }, + { + "epoch": 2.685589519650655, + "grad_norm": 0.1739282991837851, + "learning_rate": 1.658896442281624e-06, + "loss": 0.2139, + "step": 6150 + }, + { + "epoch": 2.6877729257641922, + "grad_norm": 0.1752625270758878, + "learning_rate": 1.636221416163175e-06, + "loss": 0.1918, + "step": 6155 + }, + { + "epoch": 2.6899563318777293, + "grad_norm": 0.1779820497152234, + "learning_rate": 1.6136971840612996e-06, + "loss": 0.1899, + "step": 6160 + }, + { + "epoch": 2.6921397379912664, + "grad_norm": 0.18605348108430886, + "learning_rate": 1.5913238913514317e-06, + "loss": 0.2034, + "step": 6165 + }, + { + "epoch": 2.6943231441048034, + "grad_norm": 0.18327039260922082, + "learning_rate": 1.569101682434812e-06, + "loss": 0.2151, + "step": 6170 + }, + { + "epoch": 2.6965065502183405, + "grad_norm": 0.18420649704706576, + "learning_rate": 1.5470307007375617e-06, + "loss": 0.1999, + "step": 6175 + }, + { + "epoch": 2.6986899563318776, + "grad_norm": 0.18496496761599715, + "learning_rate": 1.5251110887097547e-06, + "loss": 0.2015, + "step": 6180 + }, + { + "epoch": 2.7008733624454146, + "grad_norm": 0.18359525195308157, + "learning_rate": 1.5033429878245047e-06, + "loss": 0.1964, + "step": 6185 + }, + { + "epoch": 2.7030567685589517, + "grad_norm": 0.18877940096775106, + "learning_rate": 1.4817265385770273e-06, + "loss": 0.219, + "step": 6190 + }, + { + "epoch": 2.7052401746724892, + "grad_norm": 0.17210679893859862, + "learning_rate": 1.460261880483768e-06, + "loss": 0.2087, + "step": 6195 + }, + { + "epoch": 2.7074235807860263, + "grad_norm": 0.18497422573663475, + "learning_rate": 1.4389491520814785e-06, + "loss": 0.2073, + "step": 6200 + }, + { + "epoch": 2.7096069868995634, + "grad_norm": 0.18280651214761798, + "learning_rate": 1.417788490926328e-06, + "loss": 0.2049, + "step": 6205 + }, + { + "epoch": 2.7117903930131004, + "grad_norm": 0.21697399944996257, + "learning_rate": 1.3967800335930203e-06, + "loss": 0.1966, + "step": 6210 + }, + { + "epoch": 2.7139737991266375, + "grad_norm": 0.17484224090531814, + "learning_rate": 1.3759239156739046e-06, + "loss": 0.2004, + "step": 6215 + }, + { + "epoch": 2.7161572052401746, + "grad_norm": 0.17922630540901408, + "learning_rate": 1.355220271778107e-06, + "loss": 0.2171, + "step": 6220 + }, + { + "epoch": 2.7183406113537116, + "grad_norm": 0.1788064963960434, + "learning_rate": 1.3346692355306583e-06, + "loss": 0.21, + "step": 6225 + }, + { + "epoch": 2.720524017467249, + "grad_norm": 0.17658228011851998, + "learning_rate": 1.3142709395716325e-06, + "loss": 0.2041, + "step": 6230 + }, + { + "epoch": 2.722707423580786, + "grad_norm": 0.19747059238001105, + "learning_rate": 1.294025515555289e-06, + "loss": 0.2141, + "step": 6235 + }, + { + "epoch": 2.7248908296943233, + "grad_norm": 0.19234938030154514, + "learning_rate": 1.2739330941492211e-06, + "loss": 0.2194, + "step": 6240 + }, + { + "epoch": 2.7270742358078603, + "grad_norm": 0.18116820313254028, + "learning_rate": 1.253993805033521e-06, + "loss": 0.2026, + "step": 6245 + }, + { + "epoch": 2.7292576419213974, + "grad_norm": 0.18091742916208972, + "learning_rate": 1.2342077768999372e-06, + "loss": 0.2074, + "step": 6250 + }, + { + "epoch": 2.7314410480349345, + "grad_norm": 0.18173772524843324, + "learning_rate": 1.2145751374510462e-06, + "loss": 0.1982, + "step": 6255 + }, + { + "epoch": 2.7336244541484715, + "grad_norm": 0.18111352269398215, + "learning_rate": 1.1950960133994189e-06, + "loss": 0.2113, + "step": 6260 + }, + { + "epoch": 2.7358078602620086, + "grad_norm": 0.17797289847881081, + "learning_rate": 1.175770530466816e-06, + "loss": 0.1985, + "step": 6265 + }, + { + "epoch": 2.7379912663755457, + "grad_norm": 0.18092908617426118, + "learning_rate": 1.156598813383372e-06, + "loss": 0.2228, + "step": 6270 + }, + { + "epoch": 2.7401746724890828, + "grad_norm": 0.18444520409610834, + "learning_rate": 1.1375809858867853e-06, + "loss": 0.2114, + "step": 6275 + }, + { + "epoch": 2.74235807860262, + "grad_norm": 0.1725831612805252, + "learning_rate": 1.1187171707215228e-06, + "loss": 0.2016, + "step": 6280 + }, + { + "epoch": 2.7445414847161573, + "grad_norm": 0.1822891732232196, + "learning_rate": 1.1000074896380386e-06, + "loss": 0.2046, + "step": 6285 + }, + { + "epoch": 2.7467248908296944, + "grad_norm": 0.1837802202840671, + "learning_rate": 1.0814520633919577e-06, + "loss": 0.2063, + "step": 6290 + }, + { + "epoch": 2.7489082969432315, + "grad_norm": 0.19854848233423214, + "learning_rate": 1.063051011743335e-06, + "loss": 0.2026, + "step": 6295 + }, + { + "epoch": 2.7510917030567685, + "grad_norm": 0.1690617798747192, + "learning_rate": 1.0448044534558582e-06, + "loss": 0.2042, + "step": 6300 + }, + { + "epoch": 2.7532751091703056, + "grad_norm": 0.18680449525384113, + "learning_rate": 1.026712506296082e-06, + "loss": 0.2198, + "step": 6305 + }, + { + "epoch": 2.7554585152838427, + "grad_norm": 0.1690572517735115, + "learning_rate": 1.0087752870326817e-06, + "loss": 0.2128, + "step": 6310 + }, + { + "epoch": 2.75764192139738, + "grad_norm": 0.1741781541277763, + "learning_rate": 9.909929114356842e-07, + "loss": 0.2026, + "step": 6315 + }, + { + "epoch": 2.7598253275109172, + "grad_norm": 0.17546914742624392, + "learning_rate": 9.733654942757353e-07, + "loss": 0.1994, + "step": 6320 + }, + { + "epoch": 2.7620087336244543, + "grad_norm": 0.17660338305720158, + "learning_rate": 9.558931493233392e-07, + "loss": 0.204, + "step": 6325 + }, + { + "epoch": 2.7641921397379914, + "grad_norm": 0.1769127719144767, + "learning_rate": 9.385759893481477e-07, + "loss": 0.1996, + "step": 6330 + }, + { + "epoch": 2.7663755458515285, + "grad_norm": 0.17837387586213685, + "learning_rate": 9.214141261182196e-07, + "loss": 0.2156, + "step": 6335 + }, + { + "epoch": 2.7685589519650655, + "grad_norm": 0.18449098828135574, + "learning_rate": 9.044076703992904e-07, + "loss": 0.208, + "step": 6340 + }, + { + "epoch": 2.7707423580786026, + "grad_norm": 0.18352552163108934, + "learning_rate": 8.875567319540784e-07, + "loss": 0.2074, + "step": 6345 + }, + { + "epoch": 2.7729257641921397, + "grad_norm": 0.18503191612346148, + "learning_rate": 8.708614195415604e-07, + "loss": 0.2, + "step": 6350 + }, + { + "epoch": 2.7751091703056767, + "grad_norm": 0.1937080570900333, + "learning_rate": 8.543218409162778e-07, + "loss": 0.2082, + "step": 6355 + }, + { + "epoch": 2.777292576419214, + "grad_norm": 0.18510996106818742, + "learning_rate": 8.379381028276401e-07, + "loss": 0.2056, + "step": 6360 + }, + { + "epoch": 2.779475982532751, + "grad_norm": 0.1777159695537935, + "learning_rate": 8.217103110192276e-07, + "loss": 0.2061, + "step": 6365 + }, + { + "epoch": 2.7816593886462884, + "grad_norm": 0.1752542433085556, + "learning_rate": 8.056385702281177e-07, + "loss": 0.2007, + "step": 6370 + }, + { + "epoch": 2.7838427947598254, + "grad_norm": 0.1789778389216568, + "learning_rate": 7.89722984184213e-07, + "loss": 0.2046, + "step": 6375 + }, + { + "epoch": 2.7860262008733625, + "grad_norm": 0.17685450533969785, + "learning_rate": 7.739636556095525e-07, + "loss": 0.2065, + "step": 6380 + }, + { + "epoch": 2.7882096069868996, + "grad_norm": 0.186062026864878, + "learning_rate": 7.583606862176712e-07, + "loss": 0.2032, + "step": 6385 + }, + { + "epoch": 2.7903930131004366, + "grad_norm": 0.16820621198730876, + "learning_rate": 7.429141767129333e-07, + "loss": 0.2037, + "step": 6390 + }, + { + "epoch": 2.7925764192139737, + "grad_norm": 0.18272612053201714, + "learning_rate": 7.276242267898775e-07, + "loss": 0.2077, + "step": 6395 + }, + { + "epoch": 2.7947598253275108, + "grad_norm": 0.1818746307663275, + "learning_rate": 7.12490935132587e-07, + "loss": 0.2126, + "step": 6400 + }, + { + "epoch": 2.7969432314410483, + "grad_norm": 0.1755002936425013, + "learning_rate": 6.97514399414037e-07, + "loss": 0.2032, + "step": 6405 + }, + { + "epoch": 2.7991266375545854, + "grad_norm": 0.18013946700206887, + "learning_rate": 6.826947162954733e-07, + "loss": 0.2066, + "step": 6410 + }, + { + "epoch": 2.8013100436681224, + "grad_norm": 0.17849038284790902, + "learning_rate": 6.68031981425793e-07, + "loss": 0.2055, + "step": 6415 + }, + { + "epoch": 2.8034934497816595, + "grad_norm": 0.21383468189806515, + "learning_rate": 6.535262894409116e-07, + "loss": 0.2171, + "step": 6420 + }, + { + "epoch": 2.8056768558951966, + "grad_norm": 0.1707908400592636, + "learning_rate": 6.391777339631727e-07, + "loss": 0.1968, + "step": 6425 + }, + { + "epoch": 2.8078602620087336, + "grad_norm": 0.1928726259141998, + "learning_rate": 6.249864076007167e-07, + "loss": 0.2078, + "step": 6430 + }, + { + "epoch": 2.8100436681222707, + "grad_norm": 0.18080723535203885, + "learning_rate": 6.109524019469154e-07, + "loss": 0.2052, + "step": 6435 + }, + { + "epoch": 2.8122270742358078, + "grad_norm": 0.1785838290629133, + "learning_rate": 5.970758075797556e-07, + "loss": 0.1978, + "step": 6440 + }, + { + "epoch": 2.814410480349345, + "grad_norm": 0.17615554784359172, + "learning_rate": 5.833567140612673e-07, + "loss": 0.205, + "step": 6445 + }, + { + "epoch": 2.816593886462882, + "grad_norm": 0.18081617254199714, + "learning_rate": 5.69795209936938e-07, + "loss": 0.2085, + "step": 6450 + }, + { + "epoch": 2.818777292576419, + "grad_norm": 0.19233898776215452, + "learning_rate": 5.563913827351469e-07, + "loss": 0.2094, + "step": 6455 + }, + { + "epoch": 2.8209606986899565, + "grad_norm": 0.18538569170027117, + "learning_rate": 5.431453189666036e-07, + "loss": 0.2133, + "step": 6460 + }, + { + "epoch": 2.8231441048034935, + "grad_norm": 0.16627030997203804, + "learning_rate": 5.30057104123774e-07, + "loss": 0.1946, + "step": 6465 + }, + { + "epoch": 2.8253275109170306, + "grad_norm": 0.18546465605942317, + "learning_rate": 5.171268226803416e-07, + "loss": 0.2014, + "step": 6470 + }, + { + "epoch": 2.8275109170305677, + "grad_norm": 0.18279009388047487, + "learning_rate": 5.043545580906695e-07, + "loss": 0.2228, + "step": 6475 + }, + { + "epoch": 2.8296943231441047, + "grad_norm": 0.17398692192324666, + "learning_rate": 4.917403927892389e-07, + "loss": 0.208, + "step": 6480 + }, + { + "epoch": 2.831877729257642, + "grad_norm": 0.17416683242727254, + "learning_rate": 4.792844081901309e-07, + "loss": 0.1994, + "step": 6485 + }, + { + "epoch": 2.834061135371179, + "grad_norm": 0.177823166130051, + "learning_rate": 4.6698668468651006e-07, + "loss": 0.1939, + "step": 6490 + }, + { + "epoch": 2.8362445414847164, + "grad_norm": 0.19673421756206907, + "learning_rate": 4.548473016500854e-07, + "loss": 0.2161, + "step": 6495 + }, + { + "epoch": 2.8384279475982535, + "grad_norm": 0.1811334337251195, + "learning_rate": 4.4286633743061413e-07, + "loss": 0.2082, + "step": 6500 + }, + { + "epoch": 2.8406113537117905, + "grad_norm": 0.17564798960003072, + "learning_rate": 4.310438693553853e-07, + "loss": 0.2066, + "step": 6505 + }, + { + "epoch": 2.8427947598253276, + "grad_norm": 0.1794074378421751, + "learning_rate": 4.193799737287285e-07, + "loss": 0.2093, + "step": 6510 + }, + { + "epoch": 2.8449781659388647, + "grad_norm": 0.18373063476787263, + "learning_rate": 4.0787472583151964e-07, + "loss": 0.2235, + "step": 6515 + }, + { + "epoch": 2.8471615720524017, + "grad_norm": 0.17915751006104885, + "learning_rate": 3.965281999206899e-07, + "loss": 0.1956, + "step": 6520 + }, + { + "epoch": 2.849344978165939, + "grad_norm": 0.17324382598114016, + "learning_rate": 3.853404692287538e-07, + "loss": 0.2074, + "step": 6525 + }, + { + "epoch": 2.851528384279476, + "grad_norm": 0.18692578826449793, + "learning_rate": 3.743116059633317e-07, + "loss": 0.2174, + "step": 6530 + }, + { + "epoch": 2.853711790393013, + "grad_norm": 0.18933592310694983, + "learning_rate": 3.6344168130668097e-07, + "loss": 0.2058, + "step": 6535 + }, + { + "epoch": 2.85589519650655, + "grad_norm": 0.18309703282903245, + "learning_rate": 3.527307654152462e-07, + "loss": 0.203, + "step": 6540 + }, + { + "epoch": 2.858078602620087, + "grad_norm": 0.17336689816177772, + "learning_rate": 3.421789274192011e-07, + "loss": 0.189, + "step": 6545 + }, + { + "epoch": 2.8602620087336246, + "grad_norm": 0.17273324095781836, + "learning_rate": 3.3178623542199916e-07, + "loss": 0.2009, + "step": 6550 + }, + { + "epoch": 2.8624454148471616, + "grad_norm": 0.18054845843338757, + "learning_rate": 3.2155275649994045e-07, + "loss": 0.2125, + "step": 6555 + }, + { + "epoch": 2.8646288209606987, + "grad_norm": 0.17486255516315685, + "learning_rate": 3.1147855670172755e-07, + "loss": 0.2014, + "step": 6560 + }, + { + "epoch": 2.8668122270742358, + "grad_norm": 0.16994065735909195, + "learning_rate": 3.015637010480576e-07, + "loss": 0.1991, + "step": 6565 + }, + { + "epoch": 2.868995633187773, + "grad_norm": 0.17823602403047226, + "learning_rate": 2.918082535311839e-07, + "loss": 0.2107, + "step": 6570 + }, + { + "epoch": 2.87117903930131, + "grad_norm": 0.18534656992358386, + "learning_rate": 2.822122771145158e-07, + "loss": 0.1988, + "step": 6575 + }, + { + "epoch": 2.8733624454148474, + "grad_norm": 0.17359168867052116, + "learning_rate": 2.727758337322056e-07, + "loss": 0.2006, + "step": 6580 + }, + { + "epoch": 2.8755458515283845, + "grad_norm": 0.1818021640887144, + "learning_rate": 2.6349898428875143e-07, + "loss": 0.2107, + "step": 6585 + }, + { + "epoch": 2.8777292576419216, + "grad_norm": 0.187022315664445, + "learning_rate": 2.543817886586003e-07, + "loss": 0.2109, + "step": 6590 + }, + { + "epoch": 2.8799126637554586, + "grad_norm": 0.18065592152437351, + "learning_rate": 2.454243056857708e-07, + "loss": 0.2155, + "step": 6595 + }, + { + "epoch": 2.8820960698689957, + "grad_norm": 0.1803124984333343, + "learning_rate": 2.366265931834588e-07, + "loss": 0.2005, + "step": 6600 + }, + { + "epoch": 2.8842794759825328, + "grad_norm": 0.17978461103858198, + "learning_rate": 2.2798870793368787e-07, + "loss": 0.2135, + "step": 6605 + }, + { + "epoch": 2.88646288209607, + "grad_norm": 0.1838336256596766, + "learning_rate": 2.1951070568691222e-07, + "loss": 0.2131, + "step": 6610 + }, + { + "epoch": 2.888646288209607, + "grad_norm": 0.18102300117813305, + "learning_rate": 2.1119264116168104e-07, + "loss": 0.2064, + "step": 6615 + }, + { + "epoch": 2.890829694323144, + "grad_norm": 0.1743140768844891, + "learning_rate": 2.030345680442719e-07, + "loss": 0.1962, + "step": 6620 + }, + { + "epoch": 2.893013100436681, + "grad_norm": 0.18304778762459922, + "learning_rate": 1.950365389883524e-07, + "loss": 0.2145, + "step": 6625 + }, + { + "epoch": 2.895196506550218, + "grad_norm": 0.17457199470182785, + "learning_rate": 1.8719860561463853e-07, + "loss": 0.1986, + "step": 6630 + }, + { + "epoch": 2.8973799126637556, + "grad_norm": 0.17803797345436936, + "learning_rate": 1.7952081851055346e-07, + "loss": 0.2087, + "step": 6635 + }, + { + "epoch": 2.8995633187772927, + "grad_norm": 0.19249639309638913, + "learning_rate": 1.720032272299138e-07, + "loss": 0.208, + "step": 6640 + }, + { + "epoch": 2.9017467248908297, + "grad_norm": 0.1931146601533778, + "learning_rate": 1.6464588029260485e-07, + "loss": 0.2159, + "step": 6645 + }, + { + "epoch": 2.903930131004367, + "grad_norm": 0.17050733886371794, + "learning_rate": 1.5744882518425875e-07, + "loss": 0.2079, + "step": 6650 + }, + { + "epoch": 2.906113537117904, + "grad_norm": 0.1821050184025676, + "learning_rate": 1.5041210835596287e-07, + "loss": 0.2113, + "step": 6655 + }, + { + "epoch": 2.908296943231441, + "grad_norm": 0.1767629872679953, + "learning_rate": 1.4353577522394635e-07, + "loss": 0.2043, + "step": 6660 + }, + { + "epoch": 2.910480349344978, + "grad_norm": 0.1770612573000096, + "learning_rate": 1.3681987016929963e-07, + "loss": 0.2086, + "step": 6665 + }, + { + "epoch": 2.9126637554585155, + "grad_norm": 0.17785323864755895, + "learning_rate": 1.3026443653767762e-07, + "loss": 0.2004, + "step": 6670 + }, + { + "epoch": 2.9148471615720526, + "grad_norm": 0.17642382590921754, + "learning_rate": 1.2386951663901913e-07, + "loss": 0.206, + "step": 6675 + }, + { + "epoch": 2.9170305676855897, + "grad_norm": 0.17596251154223802, + "learning_rate": 1.1763515174729179e-07, + "loss": 0.1986, + "step": 6680 + }, + { + "epoch": 2.9192139737991267, + "grad_norm": 0.1727279406074698, + "learning_rate": 1.115613821002004e-07, + "loss": 0.2038, + "step": 6685 + }, + { + "epoch": 2.921397379912664, + "grad_norm": 0.1761644833661898, + "learning_rate": 1.0564824689894282e-07, + "loss": 0.2117, + "step": 6690 + }, + { + "epoch": 2.923580786026201, + "grad_norm": 0.1772235899771806, + "learning_rate": 9.989578430795732e-08, + "loss": 0.2014, + "step": 6695 + }, + { + "epoch": 2.925764192139738, + "grad_norm": 0.17942614591659065, + "learning_rate": 9.430403145466448e-08, + "loss": 0.2109, + "step": 6700 + }, + { + "epoch": 2.927947598253275, + "grad_norm": 0.17253943064578878, + "learning_rate": 8.887302442924517e-08, + "loss": 0.2003, + "step": 6705 + }, + { + "epoch": 2.930131004366812, + "grad_norm": 0.17852850650326985, + "learning_rate": 8.360279828438789e-08, + "loss": 0.2094, + "step": 6710 + }, + { + "epoch": 2.932314410480349, + "grad_norm": 0.17307368947339777, + "learning_rate": 7.849338703508069e-08, + "loss": 0.2064, + "step": 6715 + }, + { + "epoch": 2.934497816593886, + "grad_norm": 0.1793388736193528, + "learning_rate": 7.3544823658378e-08, + "loss": 0.1927, + "step": 6720 + }, + { + "epoch": 2.9366812227074237, + "grad_norm": 0.1803659744144063, + "learning_rate": 6.875714009319523e-08, + "loss": 0.2076, + "step": 6725 + }, + { + "epoch": 2.938864628820961, + "grad_norm": 0.17865387702225788, + "learning_rate": 6.413036724010058e-08, + "loss": 0.2003, + "step": 6730 + }, + { + "epoch": 2.941048034934498, + "grad_norm": 0.17741136523028206, + "learning_rate": 5.966453496111524e-08, + "loss": 0.2144, + "step": 6735 + }, + { + "epoch": 2.943231441048035, + "grad_norm": 0.17550555860513464, + "learning_rate": 5.5359672079524637e-08, + "loss": 0.2153, + "step": 6740 + }, + { + "epoch": 2.945414847161572, + "grad_norm": 0.17696250741970354, + "learning_rate": 5.121580637968137e-08, + "loss": 0.2054, + "step": 6745 + }, + { + "epoch": 2.947598253275109, + "grad_norm": 0.17454155560080192, + "learning_rate": 4.7232964606841454e-08, + "loss": 0.1971, + "step": 6750 + }, + { + "epoch": 2.9497816593886466, + "grad_norm": 0.17698232159770602, + "learning_rate": 4.341117246697834e-08, + "loss": 0.2085, + "step": 6755 + }, + { + "epoch": 2.9519650655021836, + "grad_norm": 0.17354434091915077, + "learning_rate": 3.9750454626624745e-08, + "loss": 0.2137, + "step": 6760 + }, + { + "epoch": 2.9541484716157207, + "grad_norm": 0.17097280016052646, + "learning_rate": 3.6250834712708846e-08, + "loss": 0.1945, + "step": 6765 + }, + { + "epoch": 2.9563318777292578, + "grad_norm": 0.17551121068500722, + "learning_rate": 3.2912335312407226e-08, + "loss": 0.1865, + "step": 6770 + }, + { + "epoch": 2.958515283842795, + "grad_norm": 0.17583508061228037, + "learning_rate": 2.9734977972989407e-08, + "loss": 0.2107, + "step": 6775 + }, + { + "epoch": 2.960698689956332, + "grad_norm": 0.17683109325722057, + "learning_rate": 2.6718783201692964e-08, + "loss": 0.2127, + "step": 6780 + }, + { + "epoch": 2.962882096069869, + "grad_norm": 0.18803887975352115, + "learning_rate": 2.3863770465576418e-08, + "loss": 0.2156, + "step": 6785 + }, + { + "epoch": 2.965065502183406, + "grad_norm": 0.1771870084804161, + "learning_rate": 2.116995819140821e-08, + "loss": 0.2182, + "step": 6790 + }, + { + "epoch": 2.967248908296943, + "grad_norm": 0.18459436255752454, + "learning_rate": 1.8637363765536264e-08, + "loss": 0.2009, + "step": 6795 + }, + { + "epoch": 2.96943231441048, + "grad_norm": 0.17264295233596325, + "learning_rate": 1.6266003533774166e-08, + "loss": 0.201, + "step": 6800 + }, + { + "epoch": 2.9716157205240172, + "grad_norm": 0.18069190701339236, + "learning_rate": 1.405589280130959e-08, + "loss": 0.214, + "step": 6805 + }, + { + "epoch": 2.9737991266375547, + "grad_norm": 0.17111956645749835, + "learning_rate": 1.2007045832590492e-08, + "loss": 0.1958, + "step": 6810 + }, + { + "epoch": 2.975982532751092, + "grad_norm": 0.17292900306530656, + "learning_rate": 1.0119475851247395e-08, + "loss": 0.1952, + "step": 6815 + }, + { + "epoch": 2.978165938864629, + "grad_norm": 0.1806661006510654, + "learning_rate": 8.393195039993473e-09, + "loss": 0.207, + "step": 6820 + }, + { + "epoch": 2.980349344978166, + "grad_norm": 0.1805502156590632, + "learning_rate": 6.82821454055238e-09, + "loss": 0.2077, + "step": 6825 + }, + { + "epoch": 2.982532751091703, + "grad_norm": 0.17317279258008123, + "learning_rate": 5.42454445359164e-09, + "loss": 0.2, + "step": 6830 + }, + { + "epoch": 2.98471615720524, + "grad_norm": 0.17476834735543392, + "learning_rate": 4.1821938386477075e-09, + "loss": 0.2028, + "step": 6835 + }, + { + "epoch": 2.986899563318777, + "grad_norm": 0.18884495387086833, + "learning_rate": 3.1011707140732317e-09, + "loss": 0.2119, + "step": 6840 + }, + { + "epoch": 2.9890829694323147, + "grad_norm": 0.17552656950302298, + "learning_rate": 2.1814820569898698e-09, + "loss": 0.2146, + "step": 6845 + }, + { + "epoch": 2.9912663755458517, + "grad_norm": 0.17245278444580078, + "learning_rate": 1.4231338032300034e-09, + "loss": 0.213, + "step": 6850 + }, + { + "epoch": 2.993449781659389, + "grad_norm": 0.17656984635938758, + "learning_rate": 8.261308473089813e-10, + "loss": 0.2045, + "step": 6855 + }, + { + "epoch": 2.995633187772926, + "grad_norm": 0.18334478736300916, + "learning_rate": 3.9047704239181247e-10, + "loss": 0.2067, + "step": 6860 + }, + { + "epoch": 2.997816593886463, + "grad_norm": 0.1826422374600176, + "learning_rate": 1.1617520026818706e-10, + "loss": 0.1943, + "step": 6865 + }, + { + "epoch": 3.0, + "grad_norm": 0.1888467433612867, + "learning_rate": 3.2270913274956573e-12, + "loss": 0.1971, + "step": 6870 + } + ], + "logging_steps": 5, + "max_steps": 6870, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 10000000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 1.440651216420864e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-6870/training_args.bin b/checkpoint-6870/training_args.bin new file mode 100644 index 0000000..5f17a22 --- /dev/null +++ b/checkpoint-6870/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19ede9ec5d95c9ba6d1490e06a35ab83d954077f265d7ad42d2b81375c33da7b +size 7736 diff --git a/checkpoint-6870/vocab.json b/checkpoint-6870/vocab.json new file mode 100644 index 0000000..6c49fc6 --- /dev/null +++ b/checkpoint-6870/vocab.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910 +size 2776833 diff --git a/config.json b/config.json new file mode 100644 index 0000000..fa2559a --- /dev/null +++ b/config.json @@ -0,0 +1,59 @@ +{ + "architectures": [ + "Qwen2ForCausalLM" + ], + "attention_dropout": 0.0, + "bos_token_id": 151643, + "eos_token_id": 151643, + "hidden_act": "silu", + "hidden_size": 3584, + "initializer_range": 0.02, + "intermediate_size": 18944, + "layer_types": [ + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention" + ], + "max_position_embeddings": 131072, + "max_window_layers": 28, + "model_type": "qwen2", + "num_attention_heads": 28, + "num_hidden_layers": 28, + "num_key_value_heads": 4, + "rms_norm_eps": 1e-06, + "rope_scaling": null, + "rope_theta": 1000000.0, + "sliding_window": null, + "tie_word_embeddings": false, + "torch_dtype": "bfloat16", + "transformers_version": "4.55.0", + "use_cache": false, + "use_mrope": false, + "use_sliding_window": false, + "vocab_size": 152064 +} diff --git a/configuration.json b/configuration.json new file mode 100644 index 0000000..bbeeda1 --- /dev/null +++ b/configuration.json @@ -0,0 +1 @@ +{"framework": "pytorch", "task": "text-generation", "allow_remote": true} \ No newline at end of file diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..7bc5028 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,6 @@ +{ + "bos_token_id": 151643, + "eos_token_id": 151643, + "max_new_tokens": 2048, + "transformers_version": "4.55.0" +} diff --git a/merges.txt b/merges.txt new file mode 100644 index 0000000..80c1a19 --- /dev/null +++ b/merges.txt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5 +size 1671853 diff --git a/model-00001-of-00004.safetensors b/model-00001-of-00004.safetensors new file mode 100644 index 0000000..85f508e --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b7eed591ff517e9f22b7e6d51bc9b93a4aea6d18841ec2070224d80d19c523a +size 4877660776 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..ecb8bc2 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2d23ea5f1df6344f99df01328398513600a92c66f133cb3d3a759d267348e75 +size 4932751008 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..5fed81c --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ceafc6ab27e5fdceea8b39ed6f307f85d847b6090a88ee188de5c09e1fd852bc +size 4330865200 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..6bf211e --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c944547784945651653accbcd4142dbc94e5ddcc47d6ea9bcf07b220b96720c9 +size 1089994880 diff --git a/model.safetensors.index.json b/model.safetensors.index.json new file mode 100644 index 0000000..098a4a8 --- /dev/null +++ b/model.safetensors.index.json @@ -0,0 +1,347 @@ +{ + "metadata": { + "total_parameters": 333312, + "total_size": 15231233024 + }, + "weight_map": { + "lm_head.weight": "model-00004-of-00004.safetensors", + "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00003-of-00004.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00003-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-00003-of-00004.safetensors", + "model.layers.18.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00003-of-00004.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.k_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.q_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.v_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00003-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.q_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.v_proj.bias": "model-00003-of-00004.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00003-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00003-of-00004.safetensors", + "model.layers.27.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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00001-of-00004.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00001-of-00004.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "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.self_attn.k_proj.bias": "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.bias": "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.bias": "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-00001-of-00004.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-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-00001-of-00004.safetensors", + "model.layers.7.self_attn.k_proj.bias": "model-00001-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.bias": "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.bias": "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.self_attn.k_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00004.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-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.self_attn.k_proj.bias": "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.bias": "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.bias": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.norm.weight": "model-00003-of-00004.safetensors" + } +} diff --git a/performance.png b/performance.png new file mode 100644 index 0000000..fea3243 Binary files /dev/null and b/performance.png differ diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000..17305b3 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,31 @@ +{ + "additional_special_tokens": [ + "<|im_start|>", + "<|im_end|>", + "<|object_ref_start|>", + "<|object_ref_end|>", + "<|box_start|>", + "<|box_end|>", + "<|quad_start|>", + "<|quad_end|>", + "<|vision_start|>", + "<|vision_end|>", + "<|vision_pad|>", + "<|image_pad|>", + "<|video_pad|>" + ], + "eos_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..51ebb3b --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa +size 11421896 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..118a723 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,209 @@ +{ + "add_bos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "151643": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151644": { + "content": "<|im_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151645": { + "content": "<|im_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151646": { + "content": "<|object_ref_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151647": { + "content": "<|object_ref_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151648": { + "content": "<|box_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151649": { + "content": "<|box_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151650": { + "content": "<|quad_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151651": { + "content": "<|quad_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151652": { + "content": "<|vision_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151653": { + "content": "<|vision_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151654": { + "content": "<|vision_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151655": { + "content": "<|image_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151656": { + "content": "<|video_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151657": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151658": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151659": { + "content": "<|fim_prefix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151660": { + "content": "<|fim_middle|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151661": { + "content": "<|fim_suffix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151662": { + "content": "<|fim_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151663": { + "content": "<|repo_name|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151664": { + "content": "<|file_sep|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "additional_special_tokens": [ + "<|im_start|>", + "<|im_end|>", + "<|object_ref_start|>", + "<|object_ref_end|>", + "<|box_start|>", + "<|box_end|>", + "<|quad_start|>", + "<|quad_end|>", + "<|vision_start|>", + "<|vision_end|>", + "<|vision_pad|>", + "<|image_pad|>", + "<|video_pad|>" + ], + "bos_token": null, + "clean_up_tokenization_spaces": false, + "eos_token": "<|endoftext|>", + "errors": "replace", + "extra_special_tokens": {}, + "model_max_length": 131072, + "pad_token": "<|endoftext|>", + "padding_side": "right", + "split_special_tokens": false, + "tokenizer_class": "Qwen2Tokenizer", + "unk_token": null, + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ system_message + '\n' }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '\nAssistant:' }}{% elif message['role'] == 'assistant' %}{{ content + '<|endoftext|>' + '\n' }}{% endif %}{% endfor %}" +} \ No newline at end of file diff --git a/train_results.json b/train_results.json new file mode 100644 index 0000000..5a41134 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 3.0, + "total_flos": 1.440651216420864e+16, + "train_loss": 0.30719185047274594, + "train_runtime": 200812.4774, + "train_samples_per_second": 1.094, + "train_steps_per_second": 0.034 +} \ No newline at end of file diff --git a/trainer_log.jsonl b/trainer_log.jsonl new file mode 100644 index 0000000..2fbc7c9 --- /dev/null +++ b/trainer_log.jsonl @@ -0,0 +1,2662 @@ +{"current_steps": 5, "total_steps": 6870, "loss": 3.5144, "lr": 2.9112081513828244e-07, "epoch": 0.002183406113537118, "percentage": 0.07, "elapsed_time": "0:01:15", "remaining_time": "1 day, 4:56:59"} +{"current_steps": 10, "total_steps": 6870, "loss": 3.3991, "lr": 6.550218340611353e-07, "epoch": 0.004366812227074236, "percentage": 0.15, "elapsed_time": "0:02:27", "remaining_time": "1 day, 4:04:08"} +{"current_steps": 15, "total_steps": 6870, "loss": 3.4485, "lr": 1.0189228529839884e-06, "epoch": 0.006550218340611353, "percentage": 0.22, "elapsed_time": "0:03:38", "remaining_time": "1 day, 3:45:48"} +{"current_steps": 20, "total_steps": 6870, "loss": 3.3308, "lr": 1.3828238719068415e-06, "epoch": 0.008733624454148471, "percentage": 0.29, "elapsed_time": "0:04:50", "remaining_time": "1 day, 3:36:26"} +{"current_steps": 25, "total_steps": 6870, "loss": 3.0579, "lr": 1.7467248908296944e-06, "epoch": 0.010917030567685589, "percentage": 0.36, "elapsed_time": "0:06:01", "remaining_time": "1 day, 3:30:32"} +{"current_steps": 30, "total_steps": 6870, "loss": 2.9493, "lr": 2.1106259097525473e-06, "epoch": 0.013100436681222707, "percentage": 0.44, "elapsed_time": "0:07:13", "remaining_time": "1 day, 3:25:52"} +{"current_steps": 35, "total_steps": 6870, "loss": 2.6072, "lr": 2.4745269286754004e-06, "epoch": 0.015283842794759825, "percentage": 0.51, "elapsed_time": "0:08:24", "remaining_time": "1 day, 3:22:25"} +{"current_steps": 40, "total_steps": 6870, "loss": 2.3367, "lr": 2.8384279475982535e-06, "epoch": 0.017467248908296942, "percentage": 0.58, "elapsed_time": "0:09:36", "remaining_time": "1 day, 3:19:23"} +{"current_steps": 45, "total_steps": 6870, "loss": 1.974, "lr": 3.2023289665211066e-06, "epoch": 0.019650655021834062, "percentage": 0.66, "elapsed_time": "0:10:47", "remaining_time": "1 day, 3:16:44"} +{"current_steps": 50, "total_steps": 6870, "loss": 1.6593, "lr": 3.5662299854439593e-06, "epoch": 0.021834061135371178, "percentage": 0.73, "elapsed_time": "0:11:58", "remaining_time": "1 day, 3:14:28"} +{"current_steps": 55, "total_steps": 6870, "loss": 1.4396, "lr": 3.930131004366813e-06, "epoch": 0.024017467248908297, "percentage": 0.8, "elapsed_time": "0:13:10", "remaining_time": "1 day, 3:12:21"} +{"current_steps": 60, "total_steps": 6870, "loss": 1.2746, "lr": 4.2940320232896655e-06, "epoch": 0.026200873362445413, "percentage": 0.87, "elapsed_time": "0:14:21", "remaining_time": "1 day, 3:10:23"} +{"current_steps": 65, "total_steps": 6870, "loss": 1.1557, "lr": 4.657933042212519e-06, "epoch": 0.028384279475982533, "percentage": 0.95, "elapsed_time": "0:15:33", "remaining_time": "1 day, 3:08:33"} +{"current_steps": 70, "total_steps": 6870, "loss": 1.1074, "lr": 5.021834061135372e-06, "epoch": 0.03056768558951965, "percentage": 1.02, "elapsed_time": "0:16:44", "remaining_time": "1 day, 3:06:51"} +{"current_steps": 75, "total_steps": 6870, "loss": 1.0167, "lr": 5.385735080058224e-06, "epoch": 0.03275109170305677, "percentage": 1.09, "elapsed_time": "0:17:56", "remaining_time": "1 day, 3:05:11"} +{"current_steps": 80, "total_steps": 6870, "loss": 0.8846, "lr": 5.749636098981077e-06, "epoch": 0.034934497816593885, "percentage": 1.16, "elapsed_time": "0:19:07", "remaining_time": "1 day, 3:03:36"} +{"current_steps": 85, "total_steps": 6870, "loss": 0.828, "lr": 6.11353711790393e-06, "epoch": 0.03711790393013101, "percentage": 1.24, "elapsed_time": "0:20:19", "remaining_time": "1 day, 3:02:08"} +{"current_steps": 90, "total_steps": 6870, "loss": 0.8344, "lr": 6.477438136826783e-06, "epoch": 0.039301310043668124, "percentage": 1.31, "elapsed_time": "0:21:30", "remaining_time": "1 day, 3:00:33"} +{"current_steps": 95, "total_steps": 6870, "loss": 0.7635, "lr": 6.841339155749636e-06, "epoch": 0.04148471615720524, "percentage": 1.38, "elapsed_time": "0:22:42", "remaining_time": "1 day, 2:59:06"} +{"current_steps": 100, "total_steps": 6870, "loss": 0.7387, "lr": 7.205240174672489e-06, "epoch": 0.043668122270742356, "percentage": 1.46, "elapsed_time": "0:23:53", "remaining_time": "1 day, 2:57:36"} +{"current_steps": 105, "total_steps": 6870, "loss": 0.7194, "lr": 7.569141193595343e-06, "epoch": 0.04585152838427948, "percentage": 1.53, "elapsed_time": "0:25:05", "remaining_time": "1 day, 2:56:20"} +{"current_steps": 110, "total_steps": 6870, "loss": 0.681, "lr": 7.933042212518195e-06, "epoch": 0.048034934497816595, "percentage": 1.6, "elapsed_time": "0:26:16", "remaining_time": "1 day, 2:54:55"} +{"current_steps": 115, "total_steps": 6870, "loss": 0.6448, "lr": 8.296943231441049e-06, "epoch": 0.05021834061135371, "percentage": 1.67, "elapsed_time": "0:27:28", "remaining_time": "1 day, 2:53:34"} +{"current_steps": 120, "total_steps": 6870, "loss": 0.6191, "lr": 8.660844250363902e-06, "epoch": 0.05240174672489083, "percentage": 1.75, "elapsed_time": "0:28:39", "remaining_time": "1 day, 2:52:13"} +{"current_steps": 125, "total_steps": 6870, "loss": 0.6076, "lr": 9.024745269286754e-06, "epoch": 0.05458515283842795, "percentage": 1.82, "elapsed_time": "0:29:51", "remaining_time": "1 day, 2:50:52"} +{"current_steps": 130, "total_steps": 6870, "loss": 0.602, "lr": 9.388646288209607e-06, "epoch": 0.056768558951965066, "percentage": 1.89, "elapsed_time": "0:31:02", "remaining_time": "1 day, 2:49:33"} +{"current_steps": 135, "total_steps": 6870, "loss": 0.6295, "lr": 9.75254730713246e-06, "epoch": 0.05895196506550218, "percentage": 1.97, "elapsed_time": "0:32:14", "remaining_time": "1 day, 2:48:16"} +{"current_steps": 140, "total_steps": 6870, "loss": 0.5524, "lr": 1.0116448326055314e-05, "epoch": 0.0611353711790393, "percentage": 2.04, "elapsed_time": "0:33:25", "remaining_time": "1 day, 2:47:00"} +{"current_steps": 145, "total_steps": 6870, "loss": 0.5627, "lr": 1.0480349344978165e-05, "epoch": 0.06331877729257641, "percentage": 2.11, "elapsed_time": "0:34:37", "remaining_time": "1 day, 2:45:44"} +{"current_steps": 150, "total_steps": 6870, "loss": 0.541, "lr": 1.084425036390102e-05, "epoch": 0.06550218340611354, "percentage": 2.18, "elapsed_time": "0:35:48", "remaining_time": "1 day, 2:44:25"} +{"current_steps": 155, "total_steps": 6870, "loss": 0.5539, "lr": 1.1208151382823872e-05, "epoch": 0.06768558951965066, "percentage": 2.26, "elapsed_time": "0:37:00", "remaining_time": "1 day, 2:43:07"} +{"current_steps": 160, "total_steps": 6870, "loss": 0.5431, "lr": 1.1572052401746727e-05, "epoch": 0.06986899563318777, "percentage": 2.33, "elapsed_time": "0:38:11", "remaining_time": "1 day, 2:41:51"} +{"current_steps": 165, "total_steps": 6870, "loss": 0.5281, "lr": 1.1935953420669578e-05, "epoch": 0.07205240174672489, "percentage": 2.4, "elapsed_time": "0:39:23", "remaining_time": "1 day, 2:40:34"} +{"current_steps": 170, "total_steps": 6870, "loss": 0.5244, "lr": 1.2299854439592432e-05, "epoch": 0.07423580786026202, "percentage": 2.47, "elapsed_time": "0:40:34", "remaining_time": "1 day, 2:39:16"} +{"current_steps": 175, "total_steps": 6870, "loss": 0.5163, "lr": 1.2663755458515283e-05, "epoch": 0.07641921397379912, "percentage": 2.55, "elapsed_time": "0:41:46", "remaining_time": "1 day, 2:38:00"} +{"current_steps": 180, "total_steps": 6870, "loss": 0.5189, "lr": 1.3027656477438139e-05, "epoch": 0.07860262008733625, "percentage": 2.62, "elapsed_time": "0:42:57", "remaining_time": "1 day, 2:36:45"} +{"current_steps": 185, "total_steps": 6870, "loss": 0.5087, "lr": 1.339155749636099e-05, "epoch": 0.08078602620087336, "percentage": 2.69, "elapsed_time": "0:44:09", "remaining_time": "1 day, 2:35:28"} +{"current_steps": 190, "total_steps": 6870, "loss": 0.4985, "lr": 1.3755458515283843e-05, "epoch": 0.08296943231441048, "percentage": 2.77, "elapsed_time": "0:45:20", "remaining_time": "1 day, 2:34:15"} +{"current_steps": 195, "total_steps": 6870, "loss": 0.4969, "lr": 1.4119359534206697e-05, "epoch": 0.0851528384279476, "percentage": 2.84, "elapsed_time": "0:46:32", "remaining_time": "1 day, 2:32:59"} +{"current_steps": 200, "total_steps": 6870, "loss": 0.5159, "lr": 1.448326055312955e-05, "epoch": 0.08733624454148471, "percentage": 2.91, "elapsed_time": "0:47:43", "remaining_time": "1 day, 2:31:43"} +{"current_steps": 205, "total_steps": 6870, "loss": 0.4902, "lr": 1.4847161572052404e-05, "epoch": 0.08951965065502183, "percentage": 2.98, "elapsed_time": "0:48:55", "remaining_time": "1 day, 2:30:28"} +{"current_steps": 210, "total_steps": 6870, "loss": 0.4899, "lr": 1.5211062590975255e-05, "epoch": 0.09170305676855896, "percentage": 3.06, "elapsed_time": "0:50:06", "remaining_time": "1 day, 2:29:12"} +{"current_steps": 215, "total_steps": 6870, "loss": 0.4644, "lr": 1.557496360989811e-05, "epoch": 0.09388646288209607, "percentage": 3.13, "elapsed_time": "0:51:18", "remaining_time": "1 day, 2:27:57"} +{"current_steps": 220, "total_steps": 6870, "loss": 0.5203, "lr": 1.5938864628820962e-05, "epoch": 0.09606986899563319, "percentage": 3.2, "elapsed_time": "0:52:29", "remaining_time": "1 day, 2:26:41"} +{"current_steps": 225, "total_steps": 6870, "loss": 0.4787, "lr": 1.6302765647743815e-05, "epoch": 0.0982532751091703, "percentage": 3.28, "elapsed_time": "0:53:41", "remaining_time": "1 day, 2:25:27"} +{"current_steps": 230, "total_steps": 6870, "loss": 0.488, "lr": 1.6666666666666667e-05, "epoch": 0.10043668122270742, "percentage": 3.35, "elapsed_time": "0:54:52", "remaining_time": "1 day, 2:24:13"} +{"current_steps": 235, "total_steps": 6870, "loss": 0.4794, "lr": 1.703056768558952e-05, "epoch": 0.10262008733624454, "percentage": 3.42, "elapsed_time": "0:56:03", "remaining_time": "1 day, 2:22:58"} +{"current_steps": 240, "total_steps": 6870, "loss": 0.4726, "lr": 1.7394468704512373e-05, "epoch": 0.10480349344978165, "percentage": 3.49, "elapsed_time": "0:57:15", "remaining_time": "1 day, 2:21:45"} +{"current_steps": 245, "total_steps": 6870, "loss": 0.4649, "lr": 1.7758369723435225e-05, "epoch": 0.10698689956331878, "percentage": 3.57, "elapsed_time": "0:58:26", "remaining_time": "1 day, 2:20:31"} +{"current_steps": 250, "total_steps": 6870, "loss": 0.459, "lr": 1.812227074235808e-05, "epoch": 0.1091703056768559, "percentage": 3.64, "elapsed_time": "0:59:38", "remaining_time": "1 day, 2:19:15"} +{"current_steps": 255, "total_steps": 6870, "loss": 0.4607, "lr": 1.848617176128093e-05, "epoch": 0.11135371179039301, "percentage": 3.71, "elapsed_time": "1:00:49", "remaining_time": "1 day, 2:18:00"} +{"current_steps": 260, "total_steps": 6870, "loss": 0.4556, "lr": 1.8850072780203784e-05, "epoch": 0.11353711790393013, "percentage": 3.78, "elapsed_time": "1:02:01", "remaining_time": "1 day, 2:16:46"} +{"current_steps": 265, "total_steps": 6870, "loss": 0.4772, "lr": 1.921397379912664e-05, "epoch": 0.11572052401746726, "percentage": 3.86, "elapsed_time": "1:03:12", "remaining_time": "1 day, 2:15:32"} +{"current_steps": 270, "total_steps": 6870, "loss": 0.4561, "lr": 1.9577874818049492e-05, "epoch": 0.11790393013100436, "percentage": 3.93, "elapsed_time": "1:04:24", "remaining_time": "1 day, 2:14:19"} +{"current_steps": 275, "total_steps": 6870, "loss": 0.4474, "lr": 1.9941775836972345e-05, "epoch": 0.12008733624454149, "percentage": 4.0, "elapsed_time": "1:05:35", "remaining_time": "1 day, 2:13:05"} +{"current_steps": 280, "total_steps": 6870, "loss": 0.4537, "lr": 2.0305676855895198e-05, "epoch": 0.1222707423580786, "percentage": 4.08, "elapsed_time": "1:06:47", "remaining_time": "1 day, 2:11:51"} +{"current_steps": 5, "total_steps": 6870, "loss": 3.5144, "lr": 2.9112081513828244e-07, "epoch": 0.002183406113537118, "percentage": 0.07, "elapsed_time": "0:01:16", "remaining_time": "1 day, 5:03:01"} +{"current_steps": 10, "total_steps": 6870, "loss": 3.399, "lr": 6.550218340611353e-07, "epoch": 0.004366812227074236, "percentage": 0.15, "elapsed_time": "0:02:27", "remaining_time": "1 day, 4:07:25"} +{"current_steps": 15, "total_steps": 6870, "loss": 3.4489, "lr": 1.0189228529839884e-06, "epoch": 0.006550218340611353, "percentage": 0.22, "elapsed_time": "0:03:39", "remaining_time": "1 day, 3:48:40"} +{"current_steps": 20, "total_steps": 6870, "loss": 3.3308, "lr": 1.3828238719068415e-06, "epoch": 0.008733624454148471, "percentage": 0.29, "elapsed_time": "0:04:50", "remaining_time": "1 day, 3:39:03"} +{"current_steps": 25, "total_steps": 6870, "loss": 3.059, "lr": 1.7467248908296944e-06, "epoch": 0.010917030567685589, "percentage": 0.36, "elapsed_time": "0:06:02", "remaining_time": "1 day, 3:32:37"} +{"current_steps": 30, "total_steps": 6870, "loss": 2.9488, "lr": 2.1106259097525473e-06, "epoch": 0.013100436681222707, "percentage": 0.44, "elapsed_time": "0:07:13", "remaining_time": "1 day, 3:27:56"} +{"current_steps": 35, "total_steps": 6870, "loss": 2.6073, "lr": 2.4745269286754004e-06, "epoch": 0.015283842794759825, "percentage": 0.51, "elapsed_time": "0:08:25", "remaining_time": "1 day, 3:24:35"} +{"current_steps": 40, "total_steps": 6870, "loss": 2.3371, "lr": 2.8384279475982535e-06, "epoch": 0.017467248908296942, "percentage": 0.58, "elapsed_time": "0:09:36", "remaining_time": "1 day, 3:21:28"} +{"current_steps": 45, "total_steps": 6870, "loss": 1.9753, "lr": 3.2023289665211066e-06, "epoch": 0.019650655021834062, "percentage": 0.66, "elapsed_time": "0:10:48", "remaining_time": "1 day, 3:19:05"} +{"current_steps": 50, "total_steps": 6870, "loss": 1.656, "lr": 3.5662299854439593e-06, "epoch": 0.021834061135371178, "percentage": 0.73, "elapsed_time": "0:11:59", "remaining_time": "1 day, 3:16:39"} +{"current_steps": 55, "total_steps": 6870, "loss": 1.4498, "lr": 3.930131004366813e-06, "epoch": 0.024017467248908297, "percentage": 0.8, "elapsed_time": "0:13:11", "remaining_time": "1 day, 3:15:16"} +{"current_steps": 60, "total_steps": 6870, "loss": 1.2845, "lr": 4.2940320232896655e-06, "epoch": 0.026200873362445413, "percentage": 0.87, "elapsed_time": "0:14:23", "remaining_time": "1 day, 3:13:10"} +{"current_steps": 65, "total_steps": 6870, "loss": 1.1601, "lr": 4.657933042212519e-06, "epoch": 0.028384279475982533, "percentage": 0.95, "elapsed_time": "0:15:34", "remaining_time": "1 day, 3:11:13"} +{"current_steps": 70, "total_steps": 6870, "loss": 1.1064, "lr": 5.021834061135372e-06, "epoch": 0.03056768558951965, "percentage": 1.02, "elapsed_time": "0:16:46", "remaining_time": "1 day, 3:09:21"} +{"current_steps": 75, "total_steps": 6870, "loss": 1.0203, "lr": 5.385735080058224e-06, "epoch": 0.03275109170305677, "percentage": 1.09, "elapsed_time": "0:17:57", "remaining_time": "1 day, 3:07:30"} +{"current_steps": 80, "total_steps": 6870, "loss": 0.8901, "lr": 5.749636098981077e-06, "epoch": 0.034934497816593885, "percentage": 1.16, "elapsed_time": "0:19:09", "remaining_time": "1 day, 3:05:48"} +{"current_steps": 85, "total_steps": 6870, "loss": 0.8333, "lr": 6.11353711790393e-06, "epoch": 0.03711790393013101, "percentage": 1.24, "elapsed_time": "0:20:20", "remaining_time": "1 day, 3:04:14"} +{"current_steps": 90, "total_steps": 6870, "loss": 0.8399, "lr": 6.477438136826783e-06, "epoch": 0.039301310043668124, "percentage": 1.31, "elapsed_time": "0:21:32", "remaining_time": "1 day, 3:02:40"} +{"current_steps": 95, "total_steps": 6870, "loss": 0.75, "lr": 6.841339155749636e-06, "epoch": 0.04148471615720524, "percentage": 1.38, "elapsed_time": "0:22:43", "remaining_time": "1 day, 3:01:05"} +{"current_steps": 100, "total_steps": 6870, "loss": 0.7335, "lr": 7.205240174672489e-06, "epoch": 0.043668122270742356, "percentage": 1.46, "elapsed_time": "0:23:55", "remaining_time": "1 day, 2:59:34"} +{"current_steps": 105, "total_steps": 6870, "loss": 0.7151, "lr": 7.569141193595343e-06, "epoch": 0.04585152838427948, "percentage": 1.53, "elapsed_time": "0:25:06", "remaining_time": "1 day, 2:58:06"} +{"current_steps": 110, "total_steps": 6870, "loss": 0.6847, "lr": 7.933042212518195e-06, "epoch": 0.048034934497816595, "percentage": 1.6, "elapsed_time": "0:26:18", "remaining_time": "1 day, 2:56:39"} +{"current_steps": 115, "total_steps": 6870, "loss": 0.646, "lr": 8.296943231441049e-06, "epoch": 0.05021834061135371, "percentage": 1.67, "elapsed_time": "0:27:29", "remaining_time": "1 day, 2:55:17"} +{"current_steps": 120, "total_steps": 6870, "loss": 0.6267, "lr": 8.660844250363902e-06, "epoch": 0.05240174672489083, "percentage": 1.75, "elapsed_time": "0:28:41", "remaining_time": "1 day, 2:53:54"} +{"current_steps": 125, "total_steps": 6870, "loss": 0.6116, "lr": 9.024745269286754e-06, "epoch": 0.05458515283842795, "percentage": 1.82, "elapsed_time": "0:29:53", "remaining_time": "1 day, 2:52:32"} +{"current_steps": 130, "total_steps": 6870, "loss": 0.6114, "lr": 9.388646288209607e-06, "epoch": 0.056768558951965066, "percentage": 1.89, "elapsed_time": "0:31:04", "remaining_time": "1 day, 2:51:11"} +{"current_steps": 135, "total_steps": 6870, "loss": 0.6372, "lr": 9.75254730713246e-06, "epoch": 0.05895196506550218, "percentage": 1.97, "elapsed_time": "0:32:16", "remaining_time": "1 day, 2:49:50"} +{"current_steps": 140, "total_steps": 6870, "loss": 0.559, "lr": 1.0116448326055314e-05, "epoch": 0.0611353711790393, "percentage": 2.04, "elapsed_time": "0:33:27", "remaining_time": "1 day, 2:48:32"} +{"current_steps": 145, "total_steps": 6870, "loss": 0.5675, "lr": 1.0480349344978165e-05, "epoch": 0.06331877729257641, "percentage": 2.11, "elapsed_time": "0:34:39", "remaining_time": "1 day, 2:47:10"} +{"current_steps": 150, "total_steps": 6870, "loss": 0.5379, "lr": 1.084425036390102e-05, "epoch": 0.06550218340611354, "percentage": 2.18, "elapsed_time": "0:35:50", "remaining_time": "1 day, 2:45:52"} +{"current_steps": 155, "total_steps": 6870, "loss": 0.5514, "lr": 1.1208151382823872e-05, "epoch": 0.06768558951965066, "percentage": 2.26, "elapsed_time": "0:37:02", "remaining_time": "1 day, 2:44:31"} +{"current_steps": 160, "total_steps": 6870, "loss": 0.5441, "lr": 1.1572052401746727e-05, "epoch": 0.06986899563318777, "percentage": 2.33, "elapsed_time": "0:38:13", "remaining_time": "1 day, 2:43:12"} +{"current_steps": 165, "total_steps": 6870, "loss": 0.5332, "lr": 1.1935953420669578e-05, "epoch": 0.07205240174672489, "percentage": 2.4, "elapsed_time": "0:39:25", "remaining_time": "1 day, 2:41:57"} +{"current_steps": 170, "total_steps": 6870, "loss": 0.5219, "lr": 1.2299854439592432e-05, "epoch": 0.07423580786026202, "percentage": 2.47, "elapsed_time": "0:40:36", "remaining_time": "1 day, 2:40:39"} +{"current_steps": 175, "total_steps": 6870, "loss": 0.5083, "lr": 1.2663755458515283e-05, "epoch": 0.07641921397379912, "percentage": 2.55, "elapsed_time": "0:41:48", "remaining_time": "1 day, 2:39:23"} +{"current_steps": 180, "total_steps": 6870, "loss": 0.5138, "lr": 1.3027656477438139e-05, "epoch": 0.07860262008733625, "percentage": 2.62, "elapsed_time": "0:42:59", "remaining_time": "1 day, 2:38:06"} +{"current_steps": 185, "total_steps": 6870, "loss": 0.5073, "lr": 1.339155749636099e-05, "epoch": 0.08078602620087336, "percentage": 2.69, "elapsed_time": "0:44:11", "remaining_time": "1 day, 2:36:48"} +{"current_steps": 190, "total_steps": 6870, "loss": 0.4968, "lr": 1.3755458515283843e-05, "epoch": 0.08296943231441048, "percentage": 2.77, "elapsed_time": "0:45:22", "remaining_time": "1 day, 2:35:32"} +{"current_steps": 195, "total_steps": 6870, "loss": 0.4953, "lr": 1.4119359534206697e-05, "epoch": 0.0851528384279476, "percentage": 2.84, "elapsed_time": "0:46:34", "remaining_time": "1 day, 2:34:16"} +{"current_steps": 200, "total_steps": 6870, "loss": 0.5135, "lr": 1.448326055312955e-05, "epoch": 0.08733624454148471, "percentage": 2.91, "elapsed_time": "0:47:46", "remaining_time": "1 day, 2:33:01"} +{"current_steps": 205, "total_steps": 6870, "loss": 0.4954, "lr": 1.4847161572052404e-05, "epoch": 0.08951965065502183, "percentage": 2.98, "elapsed_time": "0:48:57", "remaining_time": "1 day, 2:31:47"} +{"current_steps": 210, "total_steps": 6870, "loss": 0.4913, "lr": 1.5211062590975255e-05, "epoch": 0.09170305676855896, "percentage": 3.06, "elapsed_time": "0:50:09", "remaining_time": "1 day, 2:30:31"} +{"current_steps": 215, "total_steps": 6870, "loss": 0.4639, "lr": 1.557496360989811e-05, "epoch": 0.09388646288209607, "percentage": 3.13, "elapsed_time": "0:51:20", "remaining_time": "1 day, 2:29:14"} +{"current_steps": 220, "total_steps": 6870, "loss": 0.5226, "lr": 1.5938864628820962e-05, "epoch": 0.09606986899563319, "percentage": 3.2, "elapsed_time": "0:52:32", "remaining_time": "1 day, 2:27:58"} +{"current_steps": 225, "total_steps": 6870, "loss": 0.4825, "lr": 1.6302765647743815e-05, "epoch": 0.0982532751091703, "percentage": 3.28, "elapsed_time": "0:53:43", "remaining_time": "1 day, 2:26:45"} +{"current_steps": 230, "total_steps": 6870, "loss": 0.4887, "lr": 1.6666666666666667e-05, "epoch": 0.10043668122270742, "percentage": 3.35, "elapsed_time": "0:54:55", "remaining_time": "1 day, 2:25:30"} +{"current_steps": 235, "total_steps": 6870, "loss": 0.4784, "lr": 1.703056768558952e-05, "epoch": 0.10262008733624454, "percentage": 3.42, "elapsed_time": "0:56:06", "remaining_time": "1 day, 2:24:14"} +{"current_steps": 240, "total_steps": 6870, "loss": 0.4733, "lr": 1.7394468704512373e-05, "epoch": 0.10480349344978165, "percentage": 3.49, "elapsed_time": "0:57:18", "remaining_time": "1 day, 2:22:59"} +{"current_steps": 245, "total_steps": 6870, "loss": 0.4661, "lr": 1.7758369723435225e-05, "epoch": 0.10698689956331878, "percentage": 3.57, "elapsed_time": "0:58:29", "remaining_time": "1 day, 2:21:45"} +{"current_steps": 250, "total_steps": 6870, "loss": 0.4611, "lr": 1.812227074235808e-05, "epoch": 0.1091703056768559, "percentage": 3.64, "elapsed_time": "0:59:41", "remaining_time": "1 day, 2:20:31"} +{"current_steps": 255, "total_steps": 6870, "loss": 0.4623, "lr": 1.848617176128093e-05, "epoch": 0.11135371179039301, "percentage": 3.71, "elapsed_time": "1:00:52", "remaining_time": "1 day, 2:19:15"} +{"current_steps": 260, "total_steps": 6870, "loss": 0.4573, "lr": 1.8850072780203784e-05, "epoch": 0.11353711790393013, "percentage": 3.78, "elapsed_time": "1:02:04", "remaining_time": "1 day, 2:18:01"} +{"current_steps": 265, "total_steps": 6870, "loss": 0.479, "lr": 1.921397379912664e-05, "epoch": 0.11572052401746726, "percentage": 3.86, "elapsed_time": "1:03:15", "remaining_time": "1 day, 2:16:47"} +{"current_steps": 270, "total_steps": 6870, "loss": 0.453, "lr": 1.9577874818049492e-05, "epoch": 0.11790393013100436, "percentage": 3.93, "elapsed_time": "1:04:27", "remaining_time": "1 day, 2:15:33"} +{"current_steps": 275, "total_steps": 6870, "loss": 0.4475, "lr": 1.9941775836972345e-05, "epoch": 0.12008733624454149, "percentage": 4.0, "elapsed_time": "1:05:38", "remaining_time": "1 day, 2:14:19"} +{"current_steps": 280, "total_steps": 6870, "loss": 0.4546, "lr": 2.0305676855895198e-05, "epoch": 0.1222707423580786, "percentage": 4.08, "elapsed_time": "1:06:50", "remaining_time": "1 day, 2:13:07"} +{"current_steps": 285, "total_steps": 6870, "loss": 0.4308, "lr": 2.066957787481805e-05, "epoch": 0.12445414847161572, "percentage": 4.15, "elapsed_time": "1:08:01", "remaining_time": "1 day, 2:11:53"} +{"current_steps": 290, "total_steps": 6870, "loss": 0.4485, "lr": 2.1033478893740903e-05, "epoch": 0.12663755458515283, "percentage": 4.22, "elapsed_time": "1:09:13", "remaining_time": "1 day, 2:10:40"} +{"current_steps": 295, "total_steps": 6870, "loss": 0.4595, "lr": 2.1397379912663756e-05, "epoch": 0.12882096069868995, "percentage": 4.29, "elapsed_time": "1:10:24", "remaining_time": "1 day, 2:09:26"} +{"current_steps": 300, "total_steps": 6870, "loss": 0.4528, "lr": 2.176128093158661e-05, "epoch": 0.13100436681222707, "percentage": 4.37, "elapsed_time": "1:11:36", "remaining_time": "1 day, 2:08:12"} +{"current_steps": 305, "total_steps": 6870, "loss": 0.4451, "lr": 2.2125181950509464e-05, "epoch": 0.1331877729257642, "percentage": 4.44, "elapsed_time": "1:12:48", "remaining_time": "1 day, 2:07:00"} +{"current_steps": 310, "total_steps": 6870, "loss": 0.434, "lr": 2.2489082969432314e-05, "epoch": 0.13537117903930132, "percentage": 4.51, "elapsed_time": "1:13:59", "remaining_time": "1 day, 2:05:46"} +{"current_steps": 315, "total_steps": 6870, "loss": 0.4425, "lr": 2.2852983988355166e-05, "epoch": 0.13755458515283842, "percentage": 4.59, "elapsed_time": "1:15:11", "remaining_time": "1 day, 2:04:33"} +{"current_steps": 320, "total_steps": 6870, "loss": 0.4371, "lr": 2.3216885007278022e-05, "epoch": 0.13973799126637554, "percentage": 4.66, "elapsed_time": "1:16:22", "remaining_time": "1 day, 2:03:19"} +{"current_steps": 325, "total_steps": 6870, "loss": 0.4539, "lr": 2.3580786026200875e-05, "epoch": 0.14192139737991266, "percentage": 4.73, "elapsed_time": "1:17:34", "remaining_time": "1 day, 2:02:06"} +{"current_steps": 330, "total_steps": 6870, "loss": 0.4342, "lr": 2.3944687045123728e-05, "epoch": 0.14410480349344978, "percentage": 4.8, "elapsed_time": "1:18:45", "remaining_time": "1 day, 2:00:53"} +{"current_steps": 335, "total_steps": 6870, "loss": 0.414, "lr": 2.430858806404658e-05, "epoch": 0.1462882096069869, "percentage": 4.88, "elapsed_time": "1:19:57", "remaining_time": "1 day, 1:59:40"} +{"current_steps": 340, "total_steps": 6870, "loss": 0.4206, "lr": 2.4672489082969433e-05, "epoch": 0.14847161572052403, "percentage": 4.95, "elapsed_time": "1:21:08", "remaining_time": "1 day, 1:58:27"} +{"current_steps": 345, "total_steps": 6870, "loss": 0.4447, "lr": 2.503639010189229e-05, "epoch": 0.15065502183406113, "percentage": 5.02, "elapsed_time": "1:22:20", "remaining_time": "1 day, 1:57:13"} +{"current_steps": 350, "total_steps": 6870, "loss": 0.4393, "lr": 2.540029112081514e-05, "epoch": 0.15283842794759825, "percentage": 5.09, "elapsed_time": "1:23:31", "remaining_time": "1 day, 1:56:00"} +{"current_steps": 355, "total_steps": 6870, "loss": 0.4265, "lr": 2.576419213973799e-05, "epoch": 0.15502183406113537, "percentage": 5.17, "elapsed_time": "1:24:43", "remaining_time": "1 day, 1:54:49"} +{"current_steps": 360, "total_steps": 6870, "loss": 0.4309, "lr": 2.6128093158660844e-05, "epoch": 0.1572052401746725, "percentage": 5.24, "elapsed_time": "1:25:54", "remaining_time": "1 day, 1:53:35"} +{"current_steps": 365, "total_steps": 6870, "loss": 0.4206, "lr": 2.64919941775837e-05, "epoch": 0.15938864628820962, "percentage": 5.31, "elapsed_time": "1:27:06", "remaining_time": "1 day, 1:52:22"} +{"current_steps": 370, "total_steps": 6870, "loss": 0.4209, "lr": 2.6855895196506553e-05, "epoch": 0.1615720524017467, "percentage": 5.39, "elapsed_time": "1:28:17", "remaining_time": "1 day, 1:51:09"} +{"current_steps": 375, "total_steps": 6870, "loss": 0.4014, "lr": 2.7219796215429405e-05, "epoch": 0.16375545851528384, "percentage": 5.46, "elapsed_time": "1:29:29", "remaining_time": "1 day, 1:49:56"} +{"current_steps": 380, "total_steps": 6870, "loss": 0.4351, "lr": 2.7583697234352258e-05, "epoch": 0.16593886462882096, "percentage": 5.53, "elapsed_time": "1:30:40", "remaining_time": "1 day, 1:48:44"} +{"current_steps": 385, "total_steps": 6870, "loss": 0.4446, "lr": 2.794759825327511e-05, "epoch": 0.16812227074235808, "percentage": 5.6, "elapsed_time": "1:31:52", "remaining_time": "1 day, 1:47:30"} +{"current_steps": 390, "total_steps": 6870, "loss": 0.4062, "lr": 2.831149927219796e-05, "epoch": 0.1703056768558952, "percentage": 5.68, "elapsed_time": "1:33:03", "remaining_time": "1 day, 1:46:17"} +{"current_steps": 395, "total_steps": 6870, "loss": 0.449, "lr": 2.867540029112082e-05, "epoch": 0.17248908296943233, "percentage": 5.75, "elapsed_time": "1:34:15", "remaining_time": "1 day, 1:45:04"} +{"current_steps": 400, "total_steps": 6870, "loss": 0.4102, "lr": 2.9039301310043672e-05, "epoch": 0.17467248908296942, "percentage": 5.82, "elapsed_time": "1:35:26", "remaining_time": "1 day, 1:43:52"} +{"current_steps": 405, "total_steps": 6870, "loss": 0.4149, "lr": 2.940320232896652e-05, "epoch": 0.17685589519650655, "percentage": 5.9, "elapsed_time": "1:36:38", "remaining_time": "1 day, 1:42:38"} +{"current_steps": 410, "total_steps": 6870, "loss": 0.4212, "lr": 2.9767103347889374e-05, "epoch": 0.17903930131004367, "percentage": 5.97, "elapsed_time": "1:37:49", "remaining_time": "1 day, 1:41:26"} +{"current_steps": 415, "total_steps": 6870, "loss": 0.412, "lr": 3.0131004366812227e-05, "epoch": 0.1812227074235808, "percentage": 6.04, "elapsed_time": "1:39:01", "remaining_time": "1 day, 1:40:13"} +{"current_steps": 420, "total_steps": 6870, "loss": 0.4021, "lr": 3.0494905385735083e-05, "epoch": 0.18340611353711792, "percentage": 6.11, "elapsed_time": "1:40:12", "remaining_time": "1 day, 1:39:01"} +{"current_steps": 425, "total_steps": 6870, "loss": 0.4122, "lr": 3.0858806404657935e-05, "epoch": 0.185589519650655, "percentage": 6.19, "elapsed_time": "1:41:24", "remaining_time": "1 day, 1:37:48"} +{"current_steps": 430, "total_steps": 6870, "loss": 0.3988, "lr": 3.122270742358079e-05, "epoch": 0.18777292576419213, "percentage": 6.26, "elapsed_time": "1:42:36", "remaining_time": "1 day, 1:36:37"} +{"current_steps": 435, "total_steps": 6870, "loss": 0.4122, "lr": 3.158660844250364e-05, "epoch": 0.18995633187772926, "percentage": 6.33, "elapsed_time": "1:43:47", "remaining_time": "1 day, 1:35:25"} +{"current_steps": 440, "total_steps": 6870, "loss": 0.4006, "lr": 3.1950509461426493e-05, "epoch": 0.19213973799126638, "percentage": 6.4, "elapsed_time": "1:44:59", "remaining_time": "1 day, 1:34:14"} +{"current_steps": 445, "total_steps": 6870, "loss": 0.3981, "lr": 3.2314410480349346e-05, "epoch": 0.1943231441048035, "percentage": 6.48, "elapsed_time": "1:46:10", "remaining_time": "1 day, 1:33:02"} +{"current_steps": 450, "total_steps": 6870, "loss": 0.4147, "lr": 3.26783114992722e-05, "epoch": 0.1965065502183406, "percentage": 6.55, "elapsed_time": "1:47:22", "remaining_time": "1 day, 1:31:50"} +{"current_steps": 455, "total_steps": 6870, "loss": 0.4182, "lr": 3.304221251819505e-05, "epoch": 0.19868995633187772, "percentage": 6.62, "elapsed_time": "1:48:33", "remaining_time": "1 day, 1:30:37"} +{"current_steps": 460, "total_steps": 6870, "loss": 0.3963, "lr": 3.3406113537117904e-05, "epoch": 0.20087336244541484, "percentage": 6.7, "elapsed_time": "1:49:45", "remaining_time": "1 day, 1:29:25"} +{"current_steps": 465, "total_steps": 6870, "loss": 0.4063, "lr": 3.377001455604076e-05, "epoch": 0.20305676855895197, "percentage": 6.77, "elapsed_time": "1:50:56", "remaining_time": "1 day, 1:28:12"} +{"current_steps": 470, "total_steps": 6870, "loss": 0.4085, "lr": 3.413391557496361e-05, "epoch": 0.2052401746724891, "percentage": 6.84, "elapsed_time": "1:52:08", "remaining_time": "1 day, 1:26:59"} +{"current_steps": 475, "total_steps": 6870, "loss": 0.3965, "lr": 3.449781659388647e-05, "epoch": 0.2074235807860262, "percentage": 6.91, "elapsed_time": "1:53:19", "remaining_time": "1 day, 1:25:46"} +{"current_steps": 480, "total_steps": 6870, "loss": 0.4092, "lr": 3.486171761280932e-05, "epoch": 0.2096069868995633, "percentage": 6.99, "elapsed_time": "1:54:31", "remaining_time": "1 day, 1:24:34"} +{"current_steps": 485, "total_steps": 6870, "loss": 0.4016, "lr": 3.5225618631732174e-05, "epoch": 0.21179039301310043, "percentage": 7.06, "elapsed_time": "1:55:42", "remaining_time": "1 day, 1:23:22"} +{"current_steps": 490, "total_steps": 6870, "loss": 0.4088, "lr": 3.558951965065502e-05, "epoch": 0.21397379912663755, "percentage": 7.13, "elapsed_time": "1:56:54", "remaining_time": "1 day, 1:22:10"} +{"current_steps": 495, "total_steps": 6870, "loss": 0.4095, "lr": 3.595342066957787e-05, "epoch": 0.21615720524017468, "percentage": 7.21, "elapsed_time": "1:58:06", "remaining_time": "1 day, 1:20:59"} +{"current_steps": 500, "total_steps": 6870, "loss": 0.4071, "lr": 3.6317321688500726e-05, "epoch": 0.2183406113537118, "percentage": 7.28, "elapsed_time": "1:59:17", "remaining_time": "1 day, 1:19:47"} +{"current_steps": 505, "total_steps": 6870, "loss": 0.3921, "lr": 3.6681222707423585e-05, "epoch": 0.2205240174672489, "percentage": 7.35, "elapsed_time": "2:00:29", "remaining_time": "1 day, 1:18:35"} +{"current_steps": 510, "total_steps": 6870, "loss": 0.4077, "lr": 3.704512372634644e-05, "epoch": 0.22270742358078602, "percentage": 7.42, "elapsed_time": "2:01:40", "remaining_time": "1 day, 1:17:22"} +{"current_steps": 515, "total_steps": 6870, "loss": 0.412, "lr": 3.740902474526929e-05, "epoch": 0.22489082969432314, "percentage": 7.5, "elapsed_time": "2:02:52", "remaining_time": "1 day, 1:16:10"} +{"current_steps": 520, "total_steps": 6870, "loss": 0.3954, "lr": 3.777292576419214e-05, "epoch": 0.22707423580786026, "percentage": 7.57, "elapsed_time": "2:04:03", "remaining_time": "1 day, 1:14:58"} +{"current_steps": 525, "total_steps": 6870, "loss": 0.388, "lr": 3.8136826783114996e-05, "epoch": 0.2292576419213974, "percentage": 7.64, "elapsed_time": "2:05:15", "remaining_time": "1 day, 1:13:46"} +{"current_steps": 530, "total_steps": 6870, "loss": 0.4056, "lr": 3.850072780203784e-05, "epoch": 0.2314410480349345, "percentage": 7.71, "elapsed_time": "2:06:26", "remaining_time": "1 day, 1:12:34"} +{"current_steps": 535, "total_steps": 6870, "loss": 0.3948, "lr": 3.88646288209607e-05, "epoch": 0.2336244541484716, "percentage": 7.79, "elapsed_time": "2:07:38", "remaining_time": "1 day, 1:11:23"} +{"current_steps": 540, "total_steps": 6870, "loss": 0.3949, "lr": 3.9228529839883554e-05, "epoch": 0.23580786026200873, "percentage": 7.86, "elapsed_time": "2:08:49", "remaining_time": "1 day, 1:10:11"} +{"current_steps": 545, "total_steps": 6870, "loss": 0.3963, "lr": 3.9592430858806406e-05, "epoch": 0.23799126637554585, "percentage": 7.93, "elapsed_time": "2:10:01", "remaining_time": "1 day, 1:08:58"} +{"current_steps": 550, "total_steps": 6870, "loss": 0.3921, "lr": 3.995633187772926e-05, "epoch": 0.24017467248908297, "percentage": 8.01, "elapsed_time": "2:11:12", "remaining_time": "1 day, 1:07:46"} +{"current_steps": 555, "total_steps": 6870, "loss": 0.411, "lr": 4.032023289665211e-05, "epoch": 0.2423580786026201, "percentage": 8.08, "elapsed_time": "2:12:24", "remaining_time": "1 day, 1:06:33"} +{"current_steps": 560, "total_steps": 6870, "loss": 0.3841, "lr": 4.068413391557497e-05, "epoch": 0.2445414847161572, "percentage": 8.15, "elapsed_time": "2:13:35", "remaining_time": "1 day, 1:05:22"} +{"current_steps": 565, "total_steps": 6870, "loss": 0.3954, "lr": 4.104803493449782e-05, "epoch": 0.24672489082969432, "percentage": 8.22, "elapsed_time": "2:14:47", "remaining_time": "1 day, 1:04:10"} +{"current_steps": 570, "total_steps": 6870, "loss": 0.4032, "lr": 4.141193595342067e-05, "epoch": 0.24890829694323144, "percentage": 8.3, "elapsed_time": "2:15:58", "remaining_time": "1 day, 1:02:58"} +{"current_steps": 575, "total_steps": 6870, "loss": 0.3921, "lr": 4.177583697234352e-05, "epoch": 0.25109170305676853, "percentage": 8.37, "elapsed_time": "2:17:10", "remaining_time": "1 day, 1:01:46"} +{"current_steps": 580, "total_steps": 6870, "loss": 0.4158, "lr": 4.2139737991266375e-05, "epoch": 0.25327510917030566, "percentage": 8.44, "elapsed_time": "2:18:22", "remaining_time": "1 day, 1:00:34"} +{"current_steps": 585, "total_steps": 6870, "loss": 0.3816, "lr": 4.250363901018923e-05, "epoch": 0.2554585152838428, "percentage": 8.52, "elapsed_time": "2:19:33", "remaining_time": "1 day, 0:59:23"} +{"current_steps": 590, "total_steps": 6870, "loss": 0.4175, "lr": 4.286754002911209e-05, "epoch": 0.2576419213973799, "percentage": 8.59, "elapsed_time": "2:20:45", "remaining_time": "1 day, 0:58:11"} +{"current_steps": 595, "total_steps": 6870, "loss": 0.3894, "lr": 4.323144104803494e-05, "epoch": 0.259825327510917, "percentage": 8.66, "elapsed_time": "2:21:56", "remaining_time": "1 day, 0:56:59"} +{"current_steps": 600, "total_steps": 6870, "loss": 0.4103, "lr": 4.359534206695779e-05, "epoch": 0.26200873362445415, "percentage": 8.73, "elapsed_time": "2:23:08", "remaining_time": "1 day, 0:55:46"} +{"current_steps": 605, "total_steps": 6870, "loss": 0.3937, "lr": 4.395924308588064e-05, "epoch": 0.26419213973799127, "percentage": 8.81, "elapsed_time": "2:24:19", "remaining_time": "1 day, 0:54:34"} +{"current_steps": 610, "total_steps": 6870, "loss": 0.4099, "lr": 4.432314410480349e-05, "epoch": 0.2663755458515284, "percentage": 8.88, "elapsed_time": "2:25:31", "remaining_time": "1 day, 0:53:22"} +{"current_steps": 615, "total_steps": 6870, "loss": 0.3949, "lr": 4.468704512372635e-05, "epoch": 0.2685589519650655, "percentage": 8.95, "elapsed_time": "2:26:42", "remaining_time": "1 day, 0:52:11"} +{"current_steps": 620, "total_steps": 6870, "loss": 0.3943, "lr": 4.50509461426492e-05, "epoch": 0.27074235807860264, "percentage": 9.02, "elapsed_time": "2:27:54", "remaining_time": "1 day, 0:50:58"} +{"current_steps": 625, "total_steps": 6870, "loss": 0.39, "lr": 4.5414847161572056e-05, "epoch": 0.27292576419213976, "percentage": 9.1, "elapsed_time": "2:29:05", "remaining_time": "1 day, 0:49:47"} +{"current_steps": 630, "total_steps": 6870, "loss": 0.3789, "lr": 4.577874818049491e-05, "epoch": 0.27510917030567683, "percentage": 9.17, "elapsed_time": "2:30:17", "remaining_time": "1 day, 0:48:35"} +{"current_steps": 635, "total_steps": 6870, "loss": 0.3979, "lr": 4.614264919941776e-05, "epoch": 0.27729257641921395, "percentage": 9.24, "elapsed_time": "2:31:28", "remaining_time": "1 day, 0:47:22"} +{"current_steps": 640, "total_steps": 6870, "loss": 0.3949, "lr": 4.6506550218340614e-05, "epoch": 0.2794759825327511, "percentage": 9.32, "elapsed_time": "2:32:40", "remaining_time": "1 day, 0:46:10"} +{"current_steps": 645, "total_steps": 6870, "loss": 0.3997, "lr": 4.687045123726347e-05, "epoch": 0.2816593886462882, "percentage": 9.39, "elapsed_time": "2:33:51", "remaining_time": "1 day, 0:44:58"} +{"current_steps": 650, "total_steps": 6870, "loss": 0.3916, "lr": 4.723435225618632e-05, "epoch": 0.2838427947598253, "percentage": 9.46, "elapsed_time": "2:35:03", "remaining_time": "1 day, 0:43:46"} +{"current_steps": 655, "total_steps": 6870, "loss": 0.4014, "lr": 4.759825327510917e-05, "epoch": 0.28602620087336245, "percentage": 9.53, "elapsed_time": "2:36:14", "remaining_time": "1 day, 0:42:34"} +{"current_steps": 660, "total_steps": 6870, "loss": 0.4094, "lr": 4.7962154294032025e-05, "epoch": 0.28820960698689957, "percentage": 9.61, "elapsed_time": "2:37:26", "remaining_time": "1 day, 0:41:22"} +{"current_steps": 665, "total_steps": 6870, "loss": 0.389, "lr": 4.832605531295488e-05, "epoch": 0.2903930131004367, "percentage": 9.68, "elapsed_time": "2:38:37", "remaining_time": "1 day, 0:40:10"} +{"current_steps": 670, "total_steps": 6870, "loss": 0.374, "lr": 4.868995633187774e-05, "epoch": 0.2925764192139738, "percentage": 9.75, "elapsed_time": "2:39:49", "remaining_time": "1 day, 0:38:58"} +{"current_steps": 675, "total_steps": 6870, "loss": 0.4035, "lr": 4.905385735080058e-05, "epoch": 0.29475982532751094, "percentage": 9.83, "elapsed_time": "2:41:00", "remaining_time": "1 day, 0:37:46"} +{"current_steps": 680, "total_steps": 6870, "loss": 0.3863, "lr": 4.9417758369723435e-05, "epoch": 0.29694323144104806, "percentage": 9.9, "elapsed_time": "2:42:12", "remaining_time": "1 day, 0:36:34"} +{"current_steps": 685, "total_steps": 6870, "loss": 0.3862, "lr": 4.978165938864629e-05, "epoch": 0.29912663755458513, "percentage": 9.97, "elapsed_time": "2:43:24", "remaining_time": "1 day, 0:35:23"} +{"current_steps": 690, "total_steps": 6870, "loss": 0.3707, "lr": 4.999998709163553e-05, "epoch": 0.30131004366812225, "percentage": 10.04, "elapsed_time": "2:44:35", "remaining_time": "1 day, 0:34:11"} +{"current_steps": 695, "total_steps": 6870, "loss": 0.383, "lr": 4.999984187268831e-05, "epoch": 0.3034934497816594, "percentage": 10.12, "elapsed_time": "2:45:47", "remaining_time": "1 day, 0:32:59"} +{"current_steps": 700, "total_steps": 6870, "loss": 0.3769, "lr": 4.9999535300278666e-05, "epoch": 0.3056768558951965, "percentage": 10.19, "elapsed_time": "2:46:58", "remaining_time": "1 day, 0:31:46"} +{"current_steps": 705, "total_steps": 6870, "loss": 0.3932, "lr": 4.999906737638528e-05, "epoch": 0.3078602620087336, "percentage": 10.26, "elapsed_time": "2:48:10", "remaining_time": "1 day, 0:30:34"} +{"current_steps": 710, "total_steps": 6870, "loss": 0.391, "lr": 4.9998438104028226e-05, "epoch": 0.31004366812227074, "percentage": 10.33, "elapsed_time": "2:49:21", "remaining_time": "1 day, 0:29:22"} +{"current_steps": 715, "total_steps": 6870, "loss": 0.38, "lr": 4.999764748726891e-05, "epoch": 0.31222707423580787, "percentage": 10.41, "elapsed_time": "2:50:33", "remaining_time": "1 day, 0:28:10"} +{"current_steps": 720, "total_steps": 6870, "loss": 0.3833, "lr": 4.9996695531210135e-05, "epoch": 0.314410480349345, "percentage": 10.48, "elapsed_time": "2:51:44", "remaining_time": "1 day, 0:26:58"} +{"current_steps": 725, "total_steps": 6870, "loss": 0.3977, "lr": 4.9995582241996e-05, "epoch": 0.3165938864628821, "percentage": 10.55, "elapsed_time": "2:52:56", "remaining_time": "1 day, 0:25:46"} +{"current_steps": 730, "total_steps": 6870, "loss": 0.3843, "lr": 4.9994307626811865e-05, "epoch": 0.31877729257641924, "percentage": 10.63, "elapsed_time": "2:54:07", "remaining_time": "1 day, 0:24:34"} +{"current_steps": 735, "total_steps": 6870, "loss": 0.3886, "lr": 4.999287169388433e-05, "epoch": 0.32096069868995636, "percentage": 10.7, "elapsed_time": "2:55:19", "remaining_time": "1 day, 0:23:23"} +{"current_steps": 740, "total_steps": 6870, "loss": 0.4081, "lr": 4.9991274452481164e-05, "epoch": 0.3231441048034934, "percentage": 10.77, "elapsed_time": "2:56:30", "remaining_time": "1 day, 0:22:11"} +{"current_steps": 745, "total_steps": 6870, "loss": 0.3962, "lr": 4.9989515912911234e-05, "epoch": 0.32532751091703055, "percentage": 10.84, "elapsed_time": "2:57:42", "remaining_time": "1 day, 0:20:59"} +{"current_steps": 750, "total_steps": 6870, "loss": 0.4075, "lr": 4.998759608652449e-05, "epoch": 0.32751091703056767, "percentage": 10.92, "elapsed_time": "2:58:53", "remaining_time": "1 day, 0:19:47"} +{"current_steps": 755, "total_steps": 6870, "loss": 0.3973, "lr": 4.998551498571182e-05, "epoch": 0.3296943231441048, "percentage": 10.99, "elapsed_time": "3:00:05", "remaining_time": "1 day, 0:18:37"} +{"current_steps": 760, "total_steps": 6870, "loss": 0.3837, "lr": 4.9983272623905045e-05, "epoch": 0.3318777292576419, "percentage": 11.06, "elapsed_time": "3:01:17", "remaining_time": "1 day, 0:17:25"} +{"current_steps": 765, "total_steps": 6870, "loss": 0.3755, "lr": 4.9980869015576736e-05, "epoch": 0.33406113537117904, "percentage": 11.14, "elapsed_time": "3:02:28", "remaining_time": "1 day, 0:16:13"} +{"current_steps": 770, "total_steps": 6870, "loss": 0.3844, "lr": 4.997830417624024e-05, "epoch": 0.33624454148471616, "percentage": 11.21, "elapsed_time": "3:03:40", "remaining_time": "1 day, 0:15:02"} +{"current_steps": 775, "total_steps": 6870, "loss": 0.3927, "lr": 4.9975578122449485e-05, "epoch": 0.3384279475982533, "percentage": 11.28, "elapsed_time": "3:04:51", "remaining_time": "1 day, 0:13:50"} +{"current_steps": 780, "total_steps": 6870, "loss": 0.3764, "lr": 4.99726908717989e-05, "epoch": 0.3406113537117904, "percentage": 11.35, "elapsed_time": "3:06:03", "remaining_time": "1 day, 0:12:38"} +{"current_steps": 785, "total_steps": 6870, "loss": 0.3765, "lr": 4.9969642442923325e-05, "epoch": 0.34279475982532753, "percentage": 11.43, "elapsed_time": "3:07:14", "remaining_time": "1 day, 0:11:26"} +{"current_steps": 790, "total_steps": 6870, "loss": 0.3906, "lr": 4.996643285549787e-05, "epoch": 0.34497816593886466, "percentage": 11.5, "elapsed_time": "3:08:26", "remaining_time": "1 day, 0:10:14"} +{"current_steps": 795, "total_steps": 6870, "loss": 0.3735, "lr": 4.996306213023778e-05, "epoch": 0.3471615720524017, "percentage": 11.57, "elapsed_time": "3:09:37", "remaining_time": "1 day, 0:09:03"} +{"current_steps": 800, "total_steps": 6870, "loss": 0.3903, "lr": 4.995953028889833e-05, "epoch": 0.34934497816593885, "percentage": 11.64, "elapsed_time": "3:10:49", "remaining_time": "1 day, 0:07:50"} +{"current_steps": 805, "total_steps": 6870, "loss": 0.3802, "lr": 4.995583735427465e-05, "epoch": 0.35152838427947597, "percentage": 11.72, "elapsed_time": "3:12:00", "remaining_time": "1 day, 0:06:39"} +{"current_steps": 810, "total_steps": 6870, "loss": 0.3788, "lr": 4.9951983350201604e-05, "epoch": 0.3537117903930131, "percentage": 11.79, "elapsed_time": "3:13:12", "remaining_time": "1 day, 0:05:27"} +{"current_steps": 815, "total_steps": 6870, "loss": 0.3677, "lr": 4.9947968301553625e-05, "epoch": 0.3558951965065502, "percentage": 11.86, "elapsed_time": "3:14:23", "remaining_time": "1 day, 0:04:16"} +{"current_steps": 820, "total_steps": 6870, "loss": 0.363, "lr": 4.994379223424456e-05, "epoch": 0.35807860262008734, "percentage": 11.94, "elapsed_time": "3:15:35", "remaining_time": "1 day, 0:03:04"} +{"current_steps": 825, "total_steps": 6870, "loss": 0.3725, "lr": 4.99394551752275e-05, "epoch": 0.36026200873362446, "percentage": 12.01, "elapsed_time": "3:16:46", "remaining_time": "1 day, 0:01:52"} +{"current_steps": 830, "total_steps": 6870, "loss": 0.3657, "lr": 4.9934957152494585e-05, "epoch": 0.3624454148471616, "percentage": 12.08, "elapsed_time": "3:17:58", "remaining_time": "1 day, 0:00:41"} +{"current_steps": 835, "total_steps": 6870, "loss": 0.3639, "lr": 4.993029819507687e-05, "epoch": 0.3646288209606987, "percentage": 12.15, "elapsed_time": "3:19:10", "remaining_time": "23:59:29"} +{"current_steps": 840, "total_steps": 6870, "loss": 0.3797, "lr": 4.9925478333044114e-05, "epoch": 0.36681222707423583, "percentage": 12.23, "elapsed_time": "3:20:21", "remaining_time": "23:58:17"} +{"current_steps": 845, "total_steps": 6870, "loss": 0.3777, "lr": 4.992049759750455e-05, "epoch": 0.36899563318777295, "percentage": 12.3, "elapsed_time": "3:21:33", "remaining_time": "23:57:06"} +{"current_steps": 850, "total_steps": 6870, "loss": 0.372, "lr": 4.9915356020604754e-05, "epoch": 0.37117903930131, "percentage": 12.37, "elapsed_time": "3:22:44", "remaining_time": "23:55:54"} +{"current_steps": 855, "total_steps": 6870, "loss": 0.3918, "lr": 4.9910053635529376e-05, "epoch": 0.37336244541484714, "percentage": 12.45, "elapsed_time": "3:23:56", "remaining_time": "23:54:42"} +{"current_steps": 860, "total_steps": 6870, "loss": 0.3806, "lr": 4.9904590476500955e-05, "epoch": 0.37554585152838427, "percentage": 12.52, "elapsed_time": "3:25:07", "remaining_time": "23:53:31"} +{"current_steps": 865, "total_steps": 6870, "loss": 0.3699, "lr": 4.989896657877972e-05, "epoch": 0.3777292576419214, "percentage": 12.59, "elapsed_time": "3:26:19", "remaining_time": "23:52:19"} +{"current_steps": 870, "total_steps": 6870, "loss": 0.3744, "lr": 4.98931819786633e-05, "epoch": 0.3799126637554585, "percentage": 12.66, "elapsed_time": "3:27:30", "remaining_time": "23:51:07"} +{"current_steps": 875, "total_steps": 6870, "loss": 0.3661, "lr": 4.988723671348654e-05, "epoch": 0.38209606986899564, "percentage": 12.74, "elapsed_time": "3:28:42", "remaining_time": "23:49:55"} +{"current_steps": 880, "total_steps": 6870, "loss": 0.3913, "lr": 4.9881130821621255e-05, "epoch": 0.38427947598253276, "percentage": 12.81, "elapsed_time": "3:29:53", "remaining_time": "23:48:44"} +{"current_steps": 885, "total_steps": 6870, "loss": 0.3755, "lr": 4.987486434247596e-05, "epoch": 0.3864628820960699, "percentage": 12.88, "elapsed_time": "3:31:05", "remaining_time": "23:47:31"} +{"current_steps": 890, "total_steps": 6870, "loss": 0.3717, "lr": 4.9868437316495636e-05, "epoch": 0.388646288209607, "percentage": 12.95, "elapsed_time": "3:32:16", "remaining_time": "23:46:19"} +{"current_steps": 895, "total_steps": 6870, "loss": 0.3667, "lr": 4.9861849785161465e-05, "epoch": 0.39082969432314413, "percentage": 13.03, "elapsed_time": "3:33:28", "remaining_time": "23:45:08"} +{"current_steps": 900, "total_steps": 6870, "loss": 0.3721, "lr": 4.9855101790990556e-05, "epoch": 0.3930131004366812, "percentage": 13.1, "elapsed_time": "3:34:39", "remaining_time": "23:43:56"} +{"current_steps": 905, "total_steps": 6870, "loss": 0.3835, "lr": 4.9848193377535665e-05, "epoch": 0.3951965065502183, "percentage": 13.17, "elapsed_time": "3:35:51", "remaining_time": "23:42:45"} +{"current_steps": 910, "total_steps": 6870, "loss": 0.3713, "lr": 4.9841124589384944e-05, "epoch": 0.39737991266375544, "percentage": 13.25, "elapsed_time": "3:37:02", "remaining_time": "23:41:33"} +{"current_steps": 915, "total_steps": 6870, "loss": 0.3701, "lr": 4.983389547216161e-05, "epoch": 0.39956331877729256, "percentage": 13.32, "elapsed_time": "3:38:14", "remaining_time": "23:40:21"} +{"current_steps": 920, "total_steps": 6870, "loss": 0.3537, "lr": 4.9826506072523684e-05, "epoch": 0.4017467248908297, "percentage": 13.39, "elapsed_time": "3:39:26", "remaining_time": "23:39:10"} +{"current_steps": 925, "total_steps": 6870, "loss": 0.3715, "lr": 4.981895643816367e-05, "epoch": 0.4039301310043668, "percentage": 13.46, "elapsed_time": "3:40:37", "remaining_time": "23:37:58"} +{"current_steps": 930, "total_steps": 6870, "loss": 0.3687, "lr": 4.9811246617808275e-05, "epoch": 0.40611353711790393, "percentage": 13.54, "elapsed_time": "3:41:49", "remaining_time": "23:36:47"} +{"current_steps": 935, "total_steps": 6870, "loss": 0.3809, "lr": 4.980337666121806e-05, "epoch": 0.40829694323144106, "percentage": 13.61, "elapsed_time": "3:43:00", "remaining_time": "23:35:35"} +{"current_steps": 940, "total_steps": 6870, "loss": 0.3563, "lr": 4.979534661918712e-05, "epoch": 0.4104803493449782, "percentage": 13.68, "elapsed_time": "3:44:12", "remaining_time": "23:34:24"} +{"current_steps": 945, "total_steps": 6870, "loss": 0.3612, "lr": 4.978715654354281e-05, "epoch": 0.4126637554585153, "percentage": 13.76, "elapsed_time": "3:45:23", "remaining_time": "23:33:12"} +{"current_steps": 950, "total_steps": 6870, "loss": 0.3672, "lr": 4.9778806487145326e-05, "epoch": 0.4148471615720524, "percentage": 13.83, "elapsed_time": "3:46:35", "remaining_time": "23:32:00"} +{"current_steps": 955, "total_steps": 6870, "loss": 0.3655, "lr": 4.977029650388744e-05, "epoch": 0.4170305676855895, "percentage": 13.9, "elapsed_time": "3:47:46", "remaining_time": "23:30:49"} +{"current_steps": 960, "total_steps": 6870, "loss": 0.3541, "lr": 4.976162664869411e-05, "epoch": 0.4192139737991266, "percentage": 13.97, "elapsed_time": "3:48:58", "remaining_time": "23:29:37"} +{"current_steps": 965, "total_steps": 6870, "loss": 0.3587, "lr": 4.975279697752213e-05, "epoch": 0.42139737991266374, "percentage": 14.05, "elapsed_time": "3:50:10", "remaining_time": "23:28:26"} +{"current_steps": 970, "total_steps": 6870, "loss": 0.366, "lr": 4.974380754735979e-05, "epoch": 0.42358078602620086, "percentage": 14.12, "elapsed_time": "3:51:21", "remaining_time": "23:27:14"} +{"current_steps": 975, "total_steps": 6870, "loss": 0.3608, "lr": 4.9734658416226476e-05, "epoch": 0.425764192139738, "percentage": 14.19, "elapsed_time": "3:52:33", "remaining_time": "23:26:02"} +{"current_steps": 980, "total_steps": 6870, "loss": 0.3684, "lr": 4.972534964317233e-05, "epoch": 0.4279475982532751, "percentage": 14.26, "elapsed_time": "3:53:44", "remaining_time": "23:24:50"} +{"current_steps": 985, "total_steps": 6870, "loss": 0.3589, "lr": 4.9715881288277835e-05, "epoch": 0.43013100436681223, "percentage": 14.34, "elapsed_time": "3:54:56", "remaining_time": "23:23:39"} +{"current_steps": 990, "total_steps": 6870, "loss": 0.3542, "lr": 4.9706253412653446e-05, "epoch": 0.43231441048034935, "percentage": 14.41, "elapsed_time": "3:56:07", "remaining_time": "23:22:27"} +{"current_steps": 995, "total_steps": 6870, "loss": 0.3663, "lr": 4.9696466078439216e-05, "epoch": 0.4344978165938865, "percentage": 14.48, "elapsed_time": "3:57:19", "remaining_time": "23:21:15"} +{"current_steps": 1000, "total_steps": 6870, "loss": 0.3638, "lr": 4.968651934880434e-05, "epoch": 0.4366812227074236, "percentage": 14.56, "elapsed_time": "3:58:30", "remaining_time": "23:20:04"} +{"current_steps": 1005, "total_steps": 6870, "loss": 0.3592, "lr": 4.96764132879468e-05, "epoch": 0.4388646288209607, "percentage": 14.63, "elapsed_time": "3:59:42", "remaining_time": "23:18:52"} +{"current_steps": 1010, "total_steps": 6870, "loss": 0.3651, "lr": 4.9666147961092924e-05, "epoch": 0.4410480349344978, "percentage": 14.7, "elapsed_time": "4:00:53", "remaining_time": "23:17:41"} +{"current_steps": 1015, "total_steps": 6870, "loss": 0.3714, "lr": 4.9655723434496984e-05, "epoch": 0.4432314410480349, "percentage": 14.77, "elapsed_time": "4:02:05", "remaining_time": "23:16:29"} +{"current_steps": 1020, "total_steps": 6870, "loss": 0.3643, "lr": 4.9645139775440716e-05, "epoch": 0.44541484716157204, "percentage": 14.85, "elapsed_time": "4:03:17", "remaining_time": "23:15:18"} +{"current_steps": 1025, "total_steps": 6870, "loss": 0.3678, "lr": 4.9634397052232983e-05, "epoch": 0.44759825327510916, "percentage": 14.92, "elapsed_time": "4:04:28", "remaining_time": "23:14:06"} +{"current_steps": 1030, "total_steps": 6870, "loss": 0.3618, "lr": 4.9623495334209226e-05, "epoch": 0.4497816593886463, "percentage": 14.99, "elapsed_time": "4:05:40", "remaining_time": "23:12:55"} +{"current_steps": 1035, "total_steps": 6870, "loss": 0.3665, "lr": 4.96124346917311e-05, "epoch": 0.4519650655021834, "percentage": 15.07, "elapsed_time": "4:06:51", "remaining_time": "23:11:43"} +{"current_steps": 1040, "total_steps": 6870, "loss": 0.3308, "lr": 4.960121519618597e-05, "epoch": 0.45414847161572053, "percentage": 15.14, "elapsed_time": "4:08:03", "remaining_time": "23:10:31"} +{"current_steps": 1045, "total_steps": 6870, "loss": 0.3507, "lr": 4.9589836919986475e-05, "epoch": 0.45633187772925765, "percentage": 15.21, "elapsed_time": "4:09:14", "remaining_time": "23:09:20"} +{"current_steps": 1050, "total_steps": 6870, "loss": 0.3494, "lr": 4.957829993657005e-05, "epoch": 0.4585152838427948, "percentage": 15.28, "elapsed_time": "4:10:26", "remaining_time": "23:08:08"} +{"current_steps": 1055, "total_steps": 6870, "loss": 0.3595, "lr": 4.956660432039846e-05, "epoch": 0.4606986899563319, "percentage": 15.36, "elapsed_time": "4:11:37", "remaining_time": "23:06:57"} +{"current_steps": 1060, "total_steps": 6870, "loss": 0.3673, "lr": 4.955475014695729e-05, "epoch": 0.462882096069869, "percentage": 15.43, "elapsed_time": "4:12:49", "remaining_time": "23:05:45"} +{"current_steps": 1065, "total_steps": 6870, "loss": 0.3725, "lr": 4.954273749275552e-05, "epoch": 0.4650655021834061, "percentage": 15.5, "elapsed_time": "4:14:00", "remaining_time": "23:04:33"} +{"current_steps": 1070, "total_steps": 6870, "loss": 0.3481, "lr": 4.9530566435324963e-05, "epoch": 0.4672489082969432, "percentage": 15.57, "elapsed_time": "4:15:12", "remaining_time": "23:03:22"} +{"current_steps": 1075, "total_steps": 6870, "loss": 0.3705, "lr": 4.9518237053219816e-05, "epoch": 0.46943231441048033, "percentage": 15.65, "elapsed_time": "4:16:23", "remaining_time": "23:02:10"} +{"current_steps": 1080, "total_steps": 6870, "loss": 0.3474, "lr": 4.950574942601611e-05, "epoch": 0.47161572052401746, "percentage": 15.72, "elapsed_time": "4:17:35", "remaining_time": "23:00:58"} +{"current_steps": 1085, "total_steps": 6870, "loss": 0.3526, "lr": 4.9493103634311245e-05, "epoch": 0.4737991266375546, "percentage": 15.79, "elapsed_time": "4:18:47", "remaining_time": "22:59:47"} +{"current_steps": 1090, "total_steps": 6870, "loss": 0.3396, "lr": 4.948029975972341e-05, "epoch": 0.4759825327510917, "percentage": 15.87, "elapsed_time": "4:19:58", "remaining_time": "22:58:35"} +{"current_steps": 1095, "total_steps": 6870, "loss": 0.3581, "lr": 4.9467337884891127e-05, "epoch": 0.4781659388646288, "percentage": 15.94, "elapsed_time": "4:21:10", "remaining_time": "22:57:23"} +{"current_steps": 1100, "total_steps": 6870, "loss": 0.3436, "lr": 4.945421809347266e-05, "epoch": 0.48034934497816595, "percentage": 16.01, "elapsed_time": "4:22:21", "remaining_time": "22:56:11"} +{"current_steps": 1105, "total_steps": 6870, "loss": 0.3577, "lr": 4.944094047014547e-05, "epoch": 0.48253275109170307, "percentage": 16.08, "elapsed_time": "4:23:33", "remaining_time": "22:55:00"} +{"current_steps": 1110, "total_steps": 6870, "loss": 0.3363, "lr": 4.942750510060574e-05, "epoch": 0.4847161572052402, "percentage": 16.16, "elapsed_time": "4:24:44", "remaining_time": "22:53:48"} +{"current_steps": 1115, "total_steps": 6870, "loss": 0.3676, "lr": 4.9413912071567756e-05, "epoch": 0.4868995633187773, "percentage": 16.23, "elapsed_time": "4:25:56", "remaining_time": "22:52:37"} +{"current_steps": 1120, "total_steps": 6870, "loss": 0.342, "lr": 4.940016147076336e-05, "epoch": 0.4890829694323144, "percentage": 16.3, "elapsed_time": "4:27:07", "remaining_time": "22:51:25"} +{"current_steps": 1125, "total_steps": 6870, "loss": 0.361, "lr": 4.938625338694139e-05, "epoch": 0.4912663755458515, "percentage": 16.38, "elapsed_time": "4:28:19", "remaining_time": "22:50:13"} +{"current_steps": 1130, "total_steps": 6870, "loss": 0.3533, "lr": 4.937218790986712e-05, "epoch": 0.49344978165938863, "percentage": 16.45, "elapsed_time": "4:29:30", "remaining_time": "22:49:02"} +{"current_steps": 1135, "total_steps": 6870, "loss": 0.3435, "lr": 4.935796513032165e-05, "epoch": 0.49563318777292575, "percentage": 16.52, "elapsed_time": "4:30:42", "remaining_time": "22:47:50"} +{"current_steps": 1140, "total_steps": 6870, "loss": 0.3569, "lr": 4.934358514010137e-05, "epoch": 0.4978165938864629, "percentage": 16.59, "elapsed_time": "4:31:53", "remaining_time": "22:46:39"} +{"current_steps": 1145, "total_steps": 6870, "loss": 0.3542, "lr": 4.93290480320173e-05, "epoch": 0.5, "percentage": 16.67, "elapsed_time": "4:33:05", "remaining_time": "22:45:28"} +{"current_steps": 1150, "total_steps": 6870, "loss": 0.3479, "lr": 4.931435389989455e-05, "epoch": 0.5021834061135371, "percentage": 16.74, "elapsed_time": "4:34:17", "remaining_time": "22:44:16"} +{"current_steps": 1155, "total_steps": 6870, "loss": 0.3698, "lr": 4.929950283857167e-05, "epoch": 0.5043668122270742, "percentage": 16.81, "elapsed_time": "4:35:28", "remaining_time": "22:43:04"} +{"current_steps": 1160, "total_steps": 6870, "loss": 0.3648, "lr": 4.9284494943900085e-05, "epoch": 0.5065502183406113, "percentage": 16.89, "elapsed_time": "4:36:40", "remaining_time": "22:41:53"} +{"current_steps": 1165, "total_steps": 6870, "loss": 0.362, "lr": 4.926933031274344e-05, "epoch": 0.5087336244541485, "percentage": 16.96, "elapsed_time": "4:37:51", "remaining_time": "22:40:41"} +{"current_steps": 1170, "total_steps": 6870, "loss": 0.3452, "lr": 4.925400904297698e-05, "epoch": 0.5109170305676856, "percentage": 17.03, "elapsed_time": "4:39:03", "remaining_time": "22:39:29"} +{"current_steps": 1175, "total_steps": 6870, "loss": 0.3651, "lr": 4.923853123348693e-05, "epoch": 0.5131004366812227, "percentage": 17.1, "elapsed_time": "4:40:14", "remaining_time": "22:38:17"} +{"current_steps": 1180, "total_steps": 6870, "loss": 0.3539, "lr": 4.922289698416983e-05, "epoch": 0.5152838427947598, "percentage": 17.18, "elapsed_time": "4:41:26", "remaining_time": "22:37:06"} +{"current_steps": 1185, "total_steps": 6870, "loss": 0.3541, "lr": 4.920710639593196e-05, "epoch": 0.517467248908297, "percentage": 17.25, "elapsed_time": "4:42:37", "remaining_time": "22:35:54"} +{"current_steps": 1190, "total_steps": 6870, "loss": 0.3508, "lr": 4.919115957068859e-05, "epoch": 0.519650655021834, "percentage": 17.32, "elapsed_time": "4:43:49", "remaining_time": "22:34:42"} +{"current_steps": 1195, "total_steps": 6870, "loss": 0.3474, "lr": 4.917505661136339e-05, "epoch": 0.5218340611353712, "percentage": 17.39, "elapsed_time": "4:45:00", "remaining_time": "22:33:31"} +{"current_steps": 1200, "total_steps": 6870, "loss": 0.3604, "lr": 4.915879762188774e-05, "epoch": 0.5240174672489083, "percentage": 17.47, "elapsed_time": "4:46:12", "remaining_time": "22:32:19"} +{"current_steps": 1205, "total_steps": 6870, "loss": 0.3543, "lr": 4.9142382707200096e-05, "epoch": 0.5262008733624454, "percentage": 17.54, "elapsed_time": "4:47:23", "remaining_time": "22:31:07"} +{"current_steps": 1210, "total_steps": 6870, "loss": 0.3627, "lr": 4.912581197324524e-05, "epoch": 0.5283842794759825, "percentage": 17.61, "elapsed_time": "4:48:35", "remaining_time": "22:29:56"} +{"current_steps": 1215, "total_steps": 6870, "loss": 0.3468, "lr": 4.9109085526973674e-05, "epoch": 0.5305676855895196, "percentage": 17.69, "elapsed_time": "4:49:46", "remaining_time": "22:28:44"} +{"current_steps": 1220, "total_steps": 6870, "loss": 0.3586, "lr": 4.909220347634087e-05, "epoch": 0.5327510917030568, "percentage": 17.76, "elapsed_time": "4:50:58", "remaining_time": "22:27:32"} +{"current_steps": 1225, "total_steps": 6870, "loss": 0.3513, "lr": 4.907516593030662e-05, "epoch": 0.5349344978165939, "percentage": 17.83, "elapsed_time": "4:52:10", "remaining_time": "22:26:21"} +{"current_steps": 1230, "total_steps": 6870, "loss": 0.3632, "lr": 4.905797299883429e-05, "epoch": 0.537117903930131, "percentage": 17.9, "elapsed_time": "4:53:21", "remaining_time": "22:25:09"} +{"current_steps": 1235, "total_steps": 6870, "loss": 0.351, "lr": 4.904062479289015e-05, "epoch": 0.5393013100436681, "percentage": 17.98, "elapsed_time": "4:54:33", "remaining_time": "22:23:57"} +{"current_steps": 1240, "total_steps": 6870, "loss": 0.3546, "lr": 4.902312142444264e-05, "epoch": 0.5414847161572053, "percentage": 18.05, "elapsed_time": "4:55:44", "remaining_time": "22:22:46"} +{"current_steps": 1245, "total_steps": 6870, "loss": 0.3376, "lr": 4.900546300646162e-05, "epoch": 0.5436681222707423, "percentage": 18.12, "elapsed_time": "4:56:56", "remaining_time": "22:21:34"} +{"current_steps": 1250, "total_steps": 6870, "loss": 0.3565, "lr": 4.89876496529177e-05, "epoch": 0.5458515283842795, "percentage": 18.2, "elapsed_time": "4:58:07", "remaining_time": "22:20:22"} +{"current_steps": 1255, "total_steps": 6870, "loss": 0.3586, "lr": 4.896968147878145e-05, "epoch": 0.5480349344978166, "percentage": 18.27, "elapsed_time": "4:59:19", "remaining_time": "22:19:10"} +{"current_steps": 1260, "total_steps": 6870, "loss": 0.3392, "lr": 4.895155860002271e-05, "epoch": 0.5502183406113537, "percentage": 18.34, "elapsed_time": "5:00:30", "remaining_time": "22:17:58"} +{"current_steps": 1265, "total_steps": 6870, "loss": 0.3545, "lr": 4.893328113360975e-05, "epoch": 0.5524017467248908, "percentage": 18.41, "elapsed_time": "5:01:42", "remaining_time": "22:16:47"} +{"current_steps": 1270, "total_steps": 6870, "loss": 0.3447, "lr": 4.891484919750865e-05, "epoch": 0.5545851528384279, "percentage": 18.49, "elapsed_time": "5:02:53", "remaining_time": "22:15:36"} +{"current_steps": 1275, "total_steps": 6870, "loss": 0.3396, "lr": 4.889626291068241e-05, "epoch": 0.5567685589519651, "percentage": 18.56, "elapsed_time": "5:04:05", "remaining_time": "22:14:24"} +{"current_steps": 1280, "total_steps": 6870, "loss": 0.3395, "lr": 4.887752239309027e-05, "epoch": 0.5589519650655022, "percentage": 18.63, "elapsed_time": "5:05:16", "remaining_time": "22:13:12"} +{"current_steps": 1285, "total_steps": 6870, "loss": 0.3576, "lr": 4.885862776568689e-05, "epoch": 0.5611353711790393, "percentage": 18.7, "elapsed_time": "5:06:28", "remaining_time": "22:12:01"} +{"current_steps": 1290, "total_steps": 6870, "loss": 0.3499, "lr": 4.883957915042158e-05, "epoch": 0.5633187772925764, "percentage": 18.78, "elapsed_time": "5:07:39", "remaining_time": "22:10:49"} +{"current_steps": 1295, "total_steps": 6870, "loss": 0.3364, "lr": 4.882037667023752e-05, "epoch": 0.5655021834061136, "percentage": 18.85, "elapsed_time": "5:08:51", "remaining_time": "22:09:37"} +{"current_steps": 1300, "total_steps": 6870, "loss": 0.3663, "lr": 4.880102044907096e-05, "epoch": 0.5676855895196506, "percentage": 18.92, "elapsed_time": "5:10:02", "remaining_time": "22:08:26"} +{"current_steps": 1305, "total_steps": 6870, "loss": 0.3292, "lr": 4.878151061185042e-05, "epoch": 0.5698689956331878, "percentage": 19.0, "elapsed_time": "5:11:14", "remaining_time": "22:07:14"} +{"current_steps": 1310, "total_steps": 6870, "loss": 0.3484, "lr": 4.876184728449589e-05, "epoch": 0.5720524017467249, "percentage": 19.07, "elapsed_time": "5:12:25", "remaining_time": "22:06:02"} +{"current_steps": 1315, "total_steps": 6870, "loss": 0.336, "lr": 4.874203059391802e-05, "epoch": 0.574235807860262, "percentage": 19.14, "elapsed_time": "5:13:37", "remaining_time": "22:04:51"} +{"current_steps": 1320, "total_steps": 6870, "loss": 0.3553, "lr": 4.8722060668017267e-05, "epoch": 0.5764192139737991, "percentage": 19.21, "elapsed_time": "5:14:49", "remaining_time": "22:03:39"} +{"current_steps": 1325, "total_steps": 6870, "loss": 0.3613, "lr": 4.870193763568312e-05, "epoch": 0.5786026200873362, "percentage": 19.29, "elapsed_time": "5:16:00", "remaining_time": "22:02:28"} +{"current_steps": 1330, "total_steps": 6870, "loss": 0.3465, "lr": 4.8681661626793246e-05, "epoch": 0.5807860262008734, "percentage": 19.36, "elapsed_time": "5:17:12", "remaining_time": "22:01:16"} +{"current_steps": 1335, "total_steps": 6870, "loss": 0.3464, "lr": 4.866123277221264e-05, "epoch": 0.5829694323144105, "percentage": 19.43, "elapsed_time": "5:18:23", "remaining_time": "22:00:05"} +{"current_steps": 1340, "total_steps": 6870, "loss": 0.357, "lr": 4.864065120379278e-05, "epoch": 0.5851528384279476, "percentage": 19.51, "elapsed_time": "5:19:35", "remaining_time": "21:58:53"} +{"current_steps": 1345, "total_steps": 6870, "loss": 0.3593, "lr": 4.861991705437081e-05, "epoch": 0.5873362445414847, "percentage": 19.58, "elapsed_time": "5:20:46", "remaining_time": "21:57:42"} +{"current_steps": 1350, "total_steps": 6870, "loss": 0.3519, "lr": 4.859903045776866e-05, "epoch": 0.5895196506550219, "percentage": 19.65, "elapsed_time": "5:21:58", "remaining_time": "21:56:30"} +{"current_steps": 1355, "total_steps": 6870, "loss": 0.3586, "lr": 4.857799154879215e-05, "epoch": 0.5917030567685589, "percentage": 19.72, "elapsed_time": "5:23:09", "remaining_time": "21:55:18"} +{"current_steps": 1360, "total_steps": 6870, "loss": 0.342, "lr": 4.8556800463230165e-05, "epoch": 0.5938864628820961, "percentage": 19.8, "elapsed_time": "5:24:21", "remaining_time": "21:54:07"} +{"current_steps": 1365, "total_steps": 6870, "loss": 0.3272, "lr": 4.8535457337853794e-05, "epoch": 0.5960698689956332, "percentage": 19.87, "elapsed_time": "5:25:32", "remaining_time": "21:52:55"} +{"current_steps": 1370, "total_steps": 6870, "loss": 0.3452, "lr": 4.8513962310415396e-05, "epoch": 0.5982532751091703, "percentage": 19.94, "elapsed_time": "5:26:44", "remaining_time": "21:51:44"} +{"current_steps": 1375, "total_steps": 6870, "loss": 0.3526, "lr": 4.849231551964771e-05, "epoch": 0.6004366812227074, "percentage": 20.01, "elapsed_time": "5:27:55", "remaining_time": "21:50:32"} +{"current_steps": 1380, "total_steps": 6870, "loss": 0.343, "lr": 4.847051710526302e-05, "epoch": 0.6026200873362445, "percentage": 20.09, "elapsed_time": "5:29:07", "remaining_time": "21:49:20"} +{"current_steps": 1385, "total_steps": 6870, "loss": 0.3543, "lr": 4.8448567207952205e-05, "epoch": 0.6048034934497817, "percentage": 20.16, "elapsed_time": "5:30:19", "remaining_time": "21:48:09"} +{"current_steps": 1390, "total_steps": 6870, "loss": 0.3238, "lr": 4.8426465969383826e-05, "epoch": 0.6069868995633187, "percentage": 20.23, "elapsed_time": "5:31:30", "remaining_time": "21:46:57"} +{"current_steps": 1395, "total_steps": 6870, "loss": 0.3245, "lr": 4.840421353220326e-05, "epoch": 0.6091703056768559, "percentage": 20.31, "elapsed_time": "5:32:42", "remaining_time": "21:45:46"} +{"current_steps": 1400, "total_steps": 6870, "loss": 0.3331, "lr": 4.83818100400317e-05, "epoch": 0.611353711790393, "percentage": 20.38, "elapsed_time": "5:33:53", "remaining_time": "21:44:34"} +{"current_steps": 1405, "total_steps": 6870, "loss": 0.3547, "lr": 4.835925563746532e-05, "epoch": 0.6135371179039302, "percentage": 20.45, "elapsed_time": "5:35:05", "remaining_time": "21:43:22"} +{"current_steps": 1410, "total_steps": 6870, "loss": 0.3499, "lr": 4.833655047007427e-05, "epoch": 0.6157205240174672, "percentage": 20.52, "elapsed_time": "5:36:16", "remaining_time": "21:42:11"} +{"current_steps": 1415, "total_steps": 6870, "loss": 0.3351, "lr": 4.831369468440177e-05, "epoch": 0.6179039301310044, "percentage": 20.6, "elapsed_time": "5:37:28", "remaining_time": "21:40:59"} +{"current_steps": 1420, "total_steps": 6870, "loss": 0.3309, "lr": 4.8290688427963175e-05, "epoch": 0.6200873362445415, "percentage": 20.67, "elapsed_time": "5:38:39", "remaining_time": "21:39:48"} +{"current_steps": 1425, "total_steps": 6870, "loss": 0.3393, "lr": 4.8267531849244974e-05, "epoch": 0.6222707423580786, "percentage": 20.74, "elapsed_time": "5:39:51", "remaining_time": "21:38:36"} +{"current_steps": 1430, "total_steps": 6870, "loss": 0.3512, "lr": 4.8244225097703885e-05, "epoch": 0.6244541484716157, "percentage": 20.82, "elapsed_time": "5:41:02", "remaining_time": "21:37:24"} +{"current_steps": 1435, "total_steps": 6870, "loss": 0.3441, "lr": 4.8220768323765866e-05, "epoch": 0.6266375545851528, "percentage": 20.89, "elapsed_time": "5:42:14", "remaining_time": "21:36:12"} +{"current_steps": 1440, "total_steps": 6870, "loss": 0.3427, "lr": 4.819716167882513e-05, "epoch": 0.62882096069869, "percentage": 20.96, "elapsed_time": "5:43:25", "remaining_time": "21:35:01"} +{"current_steps": 1445, "total_steps": 6870, "loss": 0.3387, "lr": 4.8173405315243216e-05, "epoch": 0.631004366812227, "percentage": 21.03, "elapsed_time": "5:44:37", "remaining_time": "21:33:49"} +{"current_steps": 1450, "total_steps": 6870, "loss": 0.3372, "lr": 4.814949938634793e-05, "epoch": 0.6331877729257642, "percentage": 21.11, "elapsed_time": "5:45:48", "remaining_time": "21:32:37"} +{"current_steps": 1455, "total_steps": 6870, "loss": 0.3397, "lr": 4.8125444046432434e-05, "epoch": 0.6353711790393013, "percentage": 21.18, "elapsed_time": "5:47:00", "remaining_time": "21:31:26"} +{"current_steps": 1460, "total_steps": 6870, "loss": 0.3481, "lr": 4.810123945075421e-05, "epoch": 0.6375545851528385, "percentage": 21.25, "elapsed_time": "5:48:12", "remaining_time": "21:30:14"} +{"current_steps": 1465, "total_steps": 6870, "loss": 0.3384, "lr": 4.807688575553406e-05, "epoch": 0.6397379912663755, "percentage": 21.32, "elapsed_time": "5:49:23", "remaining_time": "21:29:03"} +{"current_steps": 1470, "total_steps": 6870, "loss": 0.3318, "lr": 4.80523831179551e-05, "epoch": 0.6419213973799127, "percentage": 21.4, "elapsed_time": "5:50:35", "remaining_time": "21:27:52"} +{"current_steps": 1475, "total_steps": 6870, "loss": 0.3379, "lr": 4.8027731696161735e-05, "epoch": 0.6441048034934498, "percentage": 21.47, "elapsed_time": "5:51:46", "remaining_time": "21:26:40"} +{"current_steps": 1480, "total_steps": 6870, "loss": 0.334, "lr": 4.8002931649258674e-05, "epoch": 0.6462882096069869, "percentage": 21.54, "elapsed_time": "5:52:58", "remaining_time": "21:25:29"} +{"current_steps": 1485, "total_steps": 6870, "loss": 0.3329, "lr": 4.7977983137309855e-05, "epoch": 0.648471615720524, "percentage": 21.62, "elapsed_time": "5:54:09", "remaining_time": "21:24:17"} +{"current_steps": 1490, "total_steps": 6870, "loss": 0.3323, "lr": 4.795288632133747e-05, "epoch": 0.6506550218340611, "percentage": 21.69, "elapsed_time": "5:55:21", "remaining_time": "21:23:06"} +{"current_steps": 1495, "total_steps": 6870, "loss": 0.3565, "lr": 4.792764136332084e-05, "epoch": 0.6528384279475983, "percentage": 21.76, "elapsed_time": "5:56:33", "remaining_time": "21:21:54"} +{"current_steps": 1500, "total_steps": 6870, "loss": 0.3304, "lr": 4.790224842619548e-05, "epoch": 0.6550218340611353, "percentage": 21.83, "elapsed_time": "5:57:44", "remaining_time": "21:20:43"} +{"current_steps": 1505, "total_steps": 6870, "loss": 0.3407, "lr": 4.787670767385194e-05, "epoch": 0.6572052401746725, "percentage": 21.91, "elapsed_time": "5:58:56", "remaining_time": "21:19:31"} +{"current_steps": 1510, "total_steps": 6870, "loss": 0.3438, "lr": 4.785101927113482e-05, "epoch": 0.6593886462882096, "percentage": 21.98, "elapsed_time": "6:00:07", "remaining_time": "21:18:20"} +{"current_steps": 1515, "total_steps": 6870, "loss": 0.35, "lr": 4.782518338384168e-05, "epoch": 0.6615720524017468, "percentage": 22.05, "elapsed_time": "6:01:19", "remaining_time": "21:17:08"} +{"current_steps": 1520, "total_steps": 6870, "loss": 0.3437, "lr": 4.779920017872197e-05, "epoch": 0.6637554585152838, "percentage": 22.13, "elapsed_time": "6:02:30", "remaining_time": "21:15:57"} +{"current_steps": 1525, "total_steps": 6870, "loss": 0.3326, "lr": 4.7773069823475945e-05, "epoch": 0.665938864628821, "percentage": 22.2, "elapsed_time": "6:03:42", "remaining_time": "21:14:45"} +{"current_steps": 1530, "total_steps": 6870, "loss": 0.3479, "lr": 4.774679248675362e-05, "epoch": 0.6681222707423581, "percentage": 22.27, "elapsed_time": "6:04:53", "remaining_time": "21:13:33"} +{"current_steps": 1535, "total_steps": 6870, "loss": 0.3316, "lr": 4.772036833815363e-05, "epoch": 0.6703056768558951, "percentage": 22.34, "elapsed_time": "6:06:05", "remaining_time": "21:12:22"} +{"current_steps": 1540, "total_steps": 6870, "loss": 0.3239, "lr": 4.7693797548222166e-05, "epoch": 0.6724890829694323, "percentage": 22.42, "elapsed_time": "6:07:16", "remaining_time": "21:11:10"} +{"current_steps": 1545, "total_steps": 6870, "loss": 0.3332, "lr": 4.766708028845188e-05, "epoch": 0.6746724890829694, "percentage": 22.49, "elapsed_time": "6:08:28", "remaining_time": "21:09:59"} +{"current_steps": 1550, "total_steps": 6870, "loss": 0.3583, "lr": 4.764021673128075e-05, "epoch": 0.6768558951965066, "percentage": 22.56, "elapsed_time": "6:09:40", "remaining_time": "21:08:47"} +{"current_steps": 1555, "total_steps": 6870, "loss": 0.3503, "lr": 4.761320705009099e-05, "epoch": 0.6790393013100436, "percentage": 22.63, "elapsed_time": "6:10:51", "remaining_time": "21:07:36"} +{"current_steps": 1560, "total_steps": 6870, "loss": 0.3543, "lr": 4.758605141920792e-05, "epoch": 0.6812227074235808, "percentage": 22.71, "elapsed_time": "6:12:03", "remaining_time": "21:06:24"} +{"current_steps": 1565, "total_steps": 6870, "loss": 0.3461, "lr": 4.755875001389887e-05, "epoch": 0.6834061135371179, "percentage": 22.78, "elapsed_time": "6:13:14", "remaining_time": "21:05:13"} +{"current_steps": 1570, "total_steps": 6870, "loss": 0.3359, "lr": 4.753130301037199e-05, "epoch": 0.6855895196506551, "percentage": 22.85, "elapsed_time": "6:14:26", "remaining_time": "21:04:01"} +{"current_steps": 1575, "total_steps": 6870, "loss": 0.3405, "lr": 4.7503710585775174e-05, "epoch": 0.6877729257641921, "percentage": 22.93, "elapsed_time": "6:15:37", "remaining_time": "21:02:49"} +{"current_steps": 1580, "total_steps": 6870, "loss": 0.3459, "lr": 4.747597291819487e-05, "epoch": 0.6899563318777293, "percentage": 23.0, "elapsed_time": "6:16:49", "remaining_time": "21:01:38"} +{"current_steps": 1585, "total_steps": 6870, "loss": 0.3387, "lr": 4.744809018665495e-05, "epoch": 0.6921397379912664, "percentage": 23.07, "elapsed_time": "6:18:00", "remaining_time": "21:00:27"} +{"current_steps": 1590, "total_steps": 6870, "loss": 0.337, "lr": 4.742006257111558e-05, "epoch": 0.6943231441048034, "percentage": 23.14, "elapsed_time": "6:19:12", "remaining_time": "20:59:15"} +{"current_steps": 1595, "total_steps": 6870, "loss": 0.3621, "lr": 4.7391890252472016e-05, "epoch": 0.6965065502183406, "percentage": 23.22, "elapsed_time": "6:20:24", "remaining_time": "20:58:03"} +{"current_steps": 1600, "total_steps": 6870, "loss": 0.3433, "lr": 4.736357341255344e-05, "epoch": 0.6986899563318777, "percentage": 23.29, "elapsed_time": "6:21:35", "remaining_time": "20:56:52"} +{"current_steps": 1605, "total_steps": 6870, "loss": 0.3334, "lr": 4.7335112234121824e-05, "epoch": 0.7008733624454149, "percentage": 23.36, "elapsed_time": "6:22:47", "remaining_time": "20:55:40"} +{"current_steps": 1610, "total_steps": 6870, "loss": 0.3406, "lr": 4.730650690087072e-05, "epoch": 0.7030567685589519, "percentage": 23.44, "elapsed_time": "6:23:58", "remaining_time": "20:54:29"} +{"current_steps": 1615, "total_steps": 6870, "loss": 0.3438, "lr": 4.727775759742408e-05, "epoch": 0.7052401746724891, "percentage": 23.51, "elapsed_time": "6:25:10", "remaining_time": "20:53:17"} +{"current_steps": 1620, "total_steps": 6870, "loss": 0.3439, "lr": 4.724886450933505e-05, "epoch": 0.7074235807860262, "percentage": 23.58, "elapsed_time": "6:26:21", "remaining_time": "20:52:06"} +{"current_steps": 1625, "total_steps": 6870, "loss": 0.3517, "lr": 4.721982782308481e-05, "epoch": 0.7096069868995634, "percentage": 23.65, "elapsed_time": "6:27:33", "remaining_time": "20:50:54"} +{"current_steps": 1630, "total_steps": 6870, "loss": 0.3457, "lr": 4.719064772608134e-05, "epoch": 0.7117903930131004, "percentage": 23.73, "elapsed_time": "6:28:44", "remaining_time": "20:49:43"} +{"current_steps": 1635, "total_steps": 6870, "loss": 0.3284, "lr": 4.7161324406658225e-05, "epoch": 0.7139737991266376, "percentage": 23.8, "elapsed_time": "6:29:56", "remaining_time": "20:48:31"} +{"current_steps": 1640, "total_steps": 6870, "loss": 0.324, "lr": 4.713185805407342e-05, "epoch": 0.7161572052401747, "percentage": 23.87, "elapsed_time": "6:31:08", "remaining_time": "20:47:20"} +{"current_steps": 1645, "total_steps": 6870, "loss": 0.3548, "lr": 4.7102248858508044e-05, "epoch": 0.7183406113537117, "percentage": 23.94, "elapsed_time": "6:32:19", "remaining_time": "20:46:08"} +{"current_steps": 1650, "total_steps": 6870, "loss": 0.3299, "lr": 4.7072497011065156e-05, "epoch": 0.7205240174672489, "percentage": 24.02, "elapsed_time": "6:33:31", "remaining_time": "20:44:57"} +{"current_steps": 1655, "total_steps": 6870, "loss": 0.3266, "lr": 4.704260270376853e-05, "epoch": 0.722707423580786, "percentage": 24.09, "elapsed_time": "6:34:42", "remaining_time": "20:43:45"} +{"current_steps": 1660, "total_steps": 6870, "loss": 0.3384, "lr": 4.7012566129561365e-05, "epoch": 0.7248908296943232, "percentage": 24.16, "elapsed_time": "6:35:54", "remaining_time": "20:42:33"} +{"current_steps": 1665, "total_steps": 6870, "loss": 0.3327, "lr": 4.6982387482305104e-05, "epoch": 0.7270742358078602, "percentage": 24.24, "elapsed_time": "6:37:05", "remaining_time": "20:41:22"} +{"current_steps": 1670, "total_steps": 6870, "loss": 0.3269, "lr": 4.6952066956778155e-05, "epoch": 0.7292576419213974, "percentage": 24.31, "elapsed_time": "6:38:17", "remaining_time": "20:40:11"} +{"current_steps": 1675, "total_steps": 6870, "loss": 0.3325, "lr": 4.692160474867463e-05, "epoch": 0.7314410480349345, "percentage": 24.38, "elapsed_time": "6:39:28", "remaining_time": "20:38:59"} +{"current_steps": 1680, "total_steps": 6870, "loss": 0.3376, "lr": 4.6891001054603056e-05, "epoch": 0.7336244541484717, "percentage": 24.45, "elapsed_time": "6:40:40", "remaining_time": "20:37:47"} +{"current_steps": 1685, "total_steps": 6870, "loss": 0.3447, "lr": 4.6860256072085196e-05, "epoch": 0.7358078602620087, "percentage": 24.53, "elapsed_time": "6:41:52", "remaining_time": "20:36:36"} +{"current_steps": 1690, "total_steps": 6870, "loss": 0.3303, "lr": 4.6829369999554664e-05, "epoch": 0.7379912663755459, "percentage": 24.6, "elapsed_time": "6:43:03", "remaining_time": "20:35:24"} +{"current_steps": 1695, "total_steps": 6870, "loss": 0.3057, "lr": 4.6798343036355704e-05, "epoch": 0.740174672489083, "percentage": 24.67, "elapsed_time": "6:44:15", "remaining_time": "20:34:13"} +{"current_steps": 1700, "total_steps": 6870, "loss": 0.3375, "lr": 4.6767175382741914e-05, "epoch": 0.74235807860262, "percentage": 24.75, "elapsed_time": "6:45:26", "remaining_time": "20:33:01"} +{"current_steps": 1705, "total_steps": 6870, "loss": 0.3538, "lr": 4.673586723987491e-05, "epoch": 0.7445414847161572, "percentage": 24.82, "elapsed_time": "6:46:38", "remaining_time": "20:31:49"} +{"current_steps": 1710, "total_steps": 6870, "loss": 0.3334, "lr": 4.670441880982307e-05, "epoch": 0.7467248908296943, "percentage": 24.89, "elapsed_time": "6:47:49", "remaining_time": "20:30:37"} +{"current_steps": 1715, "total_steps": 6870, "loss": 0.3299, "lr": 4.667283029556019e-05, "epoch": 0.7489082969432315, "percentage": 24.96, "elapsed_time": "6:49:00", "remaining_time": "20:29:26"} +{"current_steps": 1720, "total_steps": 6870, "loss": 0.3447, "lr": 4.664110190096421e-05, "epoch": 0.7510917030567685, "percentage": 25.04, "elapsed_time": "6:50:12", "remaining_time": "20:28:14"} +{"current_steps": 1725, "total_steps": 6870, "loss": 0.3413, "lr": 4.660923383081588e-05, "epoch": 0.7532751091703057, "percentage": 25.11, "elapsed_time": "6:51:23", "remaining_time": "20:27:02"} +{"current_steps": 1730, "total_steps": 6870, "loss": 0.3366, "lr": 4.657722629079744e-05, "epoch": 0.7554585152838428, "percentage": 25.18, "elapsed_time": "6:52:35", "remaining_time": "20:25:51"} +{"current_steps": 1735, "total_steps": 6870, "loss": 0.3474, "lr": 4.6545079487491285e-05, "epoch": 0.75764192139738, "percentage": 25.25, "elapsed_time": "6:53:47", "remaining_time": "20:24:39"} +{"current_steps": 1740, "total_steps": 6870, "loss": 0.335, "lr": 4.6512793628378657e-05, "epoch": 0.759825327510917, "percentage": 25.33, "elapsed_time": "6:54:58", "remaining_time": "20:23:28"} +{"current_steps": 1745, "total_steps": 6870, "loss": 0.3382, "lr": 4.648036892183828e-05, "epoch": 0.7620087336244541, "percentage": 25.4, "elapsed_time": "6:56:10", "remaining_time": "20:22:16"} +{"current_steps": 1750, "total_steps": 6870, "loss": 0.3341, "lr": 4.644780557714502e-05, "epoch": 0.7641921397379913, "percentage": 25.47, "elapsed_time": "6:57:21", "remaining_time": "20:21:04"} +{"current_steps": 1755, "total_steps": 6870, "loss": 0.3241, "lr": 4.641510380446854e-05, "epoch": 0.7663755458515283, "percentage": 25.55, "elapsed_time": "6:58:33", "remaining_time": "20:19:53"} +{"current_steps": 1760, "total_steps": 6870, "loss": 0.3374, "lr": 4.638226381487195e-05, "epoch": 0.7685589519650655, "percentage": 25.62, "elapsed_time": "6:59:44", "remaining_time": "20:18:41"} +{"current_steps": 1765, "total_steps": 6870, "loss": 0.3448, "lr": 4.634928582031043e-05, "epoch": 0.7707423580786026, "percentage": 25.69, "elapsed_time": "7:00:56", "remaining_time": "20:17:29"} +{"current_steps": 1770, "total_steps": 6870, "loss": 0.3347, "lr": 4.631617003362988e-05, "epoch": 0.7729257641921398, "percentage": 25.76, "elapsed_time": "7:02:07", "remaining_time": "20:16:17"} +{"current_steps": 1775, "total_steps": 6870, "loss": 0.3212, "lr": 4.628291666856551e-05, "epoch": 0.7751091703056768, "percentage": 25.84, "elapsed_time": "7:03:19", "remaining_time": "20:15:06"} +{"current_steps": 1780, "total_steps": 6870, "loss": 0.331, "lr": 4.6249525939740514e-05, "epoch": 0.777292576419214, "percentage": 25.91, "elapsed_time": "7:04:30", "remaining_time": "20:13:54"} +{"current_steps": 1785, "total_steps": 6870, "loss": 0.3395, "lr": 4.6215998062664636e-05, "epoch": 0.7794759825327511, "percentage": 25.98, "elapsed_time": "7:05:42", "remaining_time": "20:12:43"} +{"current_steps": 1790, "total_steps": 6870, "loss": 0.3307, "lr": 4.61823332537328e-05, "epoch": 0.7816593886462883, "percentage": 26.06, "elapsed_time": "7:06:53", "remaining_time": "20:11:31"} +{"current_steps": 1795, "total_steps": 6870, "loss": 0.3337, "lr": 4.6148531730223735e-05, "epoch": 0.7838427947598253, "percentage": 26.13, "elapsed_time": "7:08:05", "remaining_time": "20:10:20"} +{"current_steps": 1800, "total_steps": 6870, "loss": 0.3297, "lr": 4.611459371029853e-05, "epoch": 0.7860262008733624, "percentage": 26.2, "elapsed_time": "7:09:16", "remaining_time": "20:09:08"} +{"current_steps": 1805, "total_steps": 6870, "loss": 0.3272, "lr": 4.6080519412999244e-05, "epoch": 0.7882096069868996, "percentage": 26.27, "elapsed_time": "7:10:28", "remaining_time": "20:07:56"} +{"current_steps": 1810, "total_steps": 6870, "loss": 0.3359, "lr": 4.604630905824751e-05, "epoch": 0.7903930131004366, "percentage": 26.35, "elapsed_time": "7:11:39", "remaining_time": "20:06:45"} +{"current_steps": 1815, "total_steps": 6870, "loss": 0.3312, "lr": 4.601196286684309e-05, "epoch": 0.7925764192139738, "percentage": 26.42, "elapsed_time": "7:12:51", "remaining_time": "20:05:33"} +{"current_steps": 1820, "total_steps": 6870, "loss": 0.3313, "lr": 4.597748106046247e-05, "epoch": 0.7947598253275109, "percentage": 26.49, "elapsed_time": "7:14:02", "remaining_time": "20:04:22"} +{"current_steps": 1825, "total_steps": 6870, "loss": 0.3367, "lr": 4.5942863861657394e-05, "epoch": 0.7969432314410481, "percentage": 26.56, "elapsed_time": "7:15:14", "remaining_time": "20:03:10"} +{"current_steps": 1830, "total_steps": 6870, "loss": 0.3427, "lr": 4.590811149385348e-05, "epoch": 0.7991266375545851, "percentage": 26.64, "elapsed_time": "7:16:26", "remaining_time": "20:01:58"} +{"current_steps": 1835, "total_steps": 6870, "loss": 0.3256, "lr": 4.5873224181348736e-05, "epoch": 0.8013100436681223, "percentage": 26.71, "elapsed_time": "7:17:37", "remaining_time": "20:00:47"} +{"current_steps": 1840, "total_steps": 6870, "loss": 0.3314, "lr": 4.5838202149312145e-05, "epoch": 0.8034934497816594, "percentage": 26.78, "elapsed_time": "7:18:49", "remaining_time": "19:59:35"} +{"current_steps": 1845, "total_steps": 6870, "loss": 0.3319, "lr": 4.580304562378218e-05, "epoch": 0.8056768558951966, "percentage": 26.86, "elapsed_time": "7:20:00", "remaining_time": "19:58:24"} +{"current_steps": 1850, "total_steps": 6870, "loss": 0.3238, "lr": 4.576775483166536e-05, "epoch": 0.8078602620087336, "percentage": 26.93, "elapsed_time": "7:21:12", "remaining_time": "19:57:12"} +{"current_steps": 1855, "total_steps": 6870, "loss": 0.3328, "lr": 4.5732330000734775e-05, "epoch": 0.8100436681222707, "percentage": 27.0, "elapsed_time": "7:22:23", "remaining_time": "19:56:01"} +{"current_steps": 1860, "total_steps": 6870, "loss": 0.3283, "lr": 4.569677135962864e-05, "epoch": 0.8122270742358079, "percentage": 27.07, "elapsed_time": "7:23:35", "remaining_time": "19:54:49"} +{"current_steps": 1865, "total_steps": 6870, "loss": 0.3144, "lr": 4.5661079137848817e-05, "epoch": 0.8144104803493449, "percentage": 27.15, "elapsed_time": "7:24:46", "remaining_time": "19:53:38"} +{"current_steps": 1870, "total_steps": 6870, "loss": 0.3191, "lr": 4.562525356575928e-05, "epoch": 0.8165938864628821, "percentage": 27.22, "elapsed_time": "7:25:58", "remaining_time": "19:52:26"} +{"current_steps": 1875, "total_steps": 6870, "loss": 0.3311, "lr": 4.558929487458472e-05, "epoch": 0.8187772925764192, "percentage": 27.29, "elapsed_time": "7:27:09", "remaining_time": "19:51:15"} +{"current_steps": 1880, "total_steps": 6870, "loss": 0.3363, "lr": 4.5553203296408966e-05, "epoch": 0.8209606986899564, "percentage": 27.37, "elapsed_time": "7:28:21", "remaining_time": "19:50:03"} +{"current_steps": 1885, "total_steps": 6870, "loss": 0.3221, "lr": 4.551697906417353e-05, "epoch": 0.8231441048034934, "percentage": 27.44, "elapsed_time": "7:29:33", "remaining_time": "19:48:51"} +{"current_steps": 1890, "total_steps": 6870, "loss": 0.3373, "lr": 4.5480622411676105e-05, "epoch": 0.8253275109170306, "percentage": 27.51, "elapsed_time": "7:30:44", "remaining_time": "19:47:40"} +{"current_steps": 1895, "total_steps": 6870, "loss": 0.3396, "lr": 4.5444133573569056e-05, "epoch": 0.8275109170305677, "percentage": 27.58, "elapsed_time": "7:31:56", "remaining_time": "19:46:28"} +{"current_steps": 1900, "total_steps": 6870, "loss": 0.3424, "lr": 4.540751278535788e-05, "epoch": 0.8296943231441049, "percentage": 27.66, "elapsed_time": "7:33:07", "remaining_time": "19:45:17"} +{"current_steps": 1905, "total_steps": 6870, "loss": 0.3228, "lr": 4.5370760283399714e-05, "epoch": 0.8318777292576419, "percentage": 27.73, "elapsed_time": "7:34:19", "remaining_time": "19:44:05"} +{"current_steps": 1910, "total_steps": 6870, "loss": 0.3249, "lr": 4.533387630490179e-05, "epoch": 0.834061135371179, "percentage": 27.8, "elapsed_time": "7:35:30", "remaining_time": "19:42:54"} +{"current_steps": 1915, "total_steps": 6870, "loss": 0.3358, "lr": 4.529686108791993e-05, "epoch": 0.8362445414847162, "percentage": 27.87, "elapsed_time": "7:36:42", "remaining_time": "19:41:42"} +{"current_steps": 1920, "total_steps": 6870, "loss": 0.3257, "lr": 4.525971487135697e-05, "epoch": 0.8384279475982532, "percentage": 27.95, "elapsed_time": "7:37:53", "remaining_time": "19:40:30"} +{"current_steps": 1925, "total_steps": 6870, "loss": 0.3311, "lr": 4.522243789496125e-05, "epoch": 0.8406113537117904, "percentage": 28.02, "elapsed_time": "7:39:05", "remaining_time": "19:39:19"} +{"current_steps": 1930, "total_steps": 6870, "loss": 0.329, "lr": 4.518503039932508e-05, "epoch": 0.8427947598253275, "percentage": 28.09, "elapsed_time": "7:40:16", "remaining_time": "19:38:07"} +{"current_steps": 1935, "total_steps": 6870, "loss": 0.3342, "lr": 4.5147492625883133e-05, "epoch": 0.8449781659388647, "percentage": 28.17, "elapsed_time": "7:41:28", "remaining_time": "19:36:56"} +{"current_steps": 1940, "total_steps": 6870, "loss": 0.3352, "lr": 4.510982481691093e-05, "epoch": 0.8471615720524017, "percentage": 28.24, "elapsed_time": "7:42:40", "remaining_time": "19:35:44"} +{"current_steps": 1945, "total_steps": 6870, "loss": 0.3279, "lr": 4.5072027215523256e-05, "epoch": 0.8493449781659389, "percentage": 28.31, "elapsed_time": "7:43:51", "remaining_time": "19:34:33"} +{"current_steps": 1950, "total_steps": 6870, "loss": 0.3141, "lr": 4.503410006567262e-05, "epoch": 0.851528384279476, "percentage": 28.38, "elapsed_time": "7:45:03", "remaining_time": "19:33:22"} +{"current_steps": 1955, "total_steps": 6870, "loss": 0.3381, "lr": 4.499604361214764e-05, "epoch": 0.8537117903930131, "percentage": 28.46, "elapsed_time": "7:46:14", "remaining_time": "19:32:10"} +{"current_steps": 1960, "total_steps": 6870, "loss": 0.3309, "lr": 4.495785810057151e-05, "epoch": 0.8558951965065502, "percentage": 28.53, "elapsed_time": "7:47:26", "remaining_time": "19:30:58"} +{"current_steps": 1965, "total_steps": 6870, "loss": 0.3152, "lr": 4.4919543777400354e-05, "epoch": 0.8580786026200873, "percentage": 28.6, "elapsed_time": "7:48:37", "remaining_time": "19:29:47"} +{"current_steps": 1970, "total_steps": 6870, "loss": 0.3372, "lr": 4.488110088992168e-05, "epoch": 0.8602620087336245, "percentage": 28.68, "elapsed_time": "7:49:49", "remaining_time": "19:28:36"} +{"current_steps": 1975, "total_steps": 6870, "loss": 0.3416, "lr": 4.4842529686252774e-05, "epoch": 0.8624454148471615, "percentage": 28.75, "elapsed_time": "7:51:01", "remaining_time": "19:27:24"} +{"current_steps": 1980, "total_steps": 6870, "loss": 0.3322, "lr": 4.4803830415339124e-05, "epoch": 0.8646288209606987, "percentage": 28.82, "elapsed_time": "7:52:12", "remaining_time": "19:26:13"} +{"current_steps": 1985, "total_steps": 6870, "loss": 0.3127, "lr": 4.476500332695275e-05, "epoch": 0.8668122270742358, "percentage": 28.89, "elapsed_time": "7:53:24", "remaining_time": "19:25:01"} +{"current_steps": 1990, "total_steps": 6870, "loss": 0.3263, "lr": 4.472604867169063e-05, "epoch": 0.868995633187773, "percentage": 28.97, "elapsed_time": "7:54:35", "remaining_time": "19:23:49"} +{"current_steps": 1995, "total_steps": 6870, "loss": 0.3326, "lr": 4.468696670097312e-05, "epoch": 0.87117903930131, "percentage": 29.04, "elapsed_time": "7:55:47", "remaining_time": "19:22:38"} +{"current_steps": 2000, "total_steps": 6870, "loss": 0.3393, "lr": 4.464775766704225e-05, "epoch": 0.8733624454148472, "percentage": 29.11, "elapsed_time": "7:56:58", "remaining_time": "19:21:26"} +{"current_steps": 2005, "total_steps": 6870, "loss": 0.3236, "lr": 4.4608421822960165e-05, "epoch": 0.8755458515283843, "percentage": 29.18, "elapsed_time": "7:58:10", "remaining_time": "19:20:14"} +{"current_steps": 2010, "total_steps": 6870, "loss": 0.3118, "lr": 4.456895942260745e-05, "epoch": 0.8777292576419214, "percentage": 29.26, "elapsed_time": "7:59:21", "remaining_time": "19:19:03"} +{"current_steps": 2015, "total_steps": 6870, "loss": 0.3332, "lr": 4.4529370720681516e-05, "epoch": 0.8799126637554585, "percentage": 29.33, "elapsed_time": "8:00:33", "remaining_time": "19:17:51"} +{"current_steps": 2020, "total_steps": 6870, "loss": 0.3238, "lr": 4.4489655972694946e-05, "epoch": 0.8820960698689956, "percentage": 29.4, "elapsed_time": "8:01:44", "remaining_time": "19:16:39"} +{"current_steps": 2025, "total_steps": 6870, "loss": 0.3269, "lr": 4.444981543497384e-05, "epoch": 0.8842794759825328, "percentage": 29.48, "elapsed_time": "8:02:56", "remaining_time": "19:15:28"} +{"current_steps": 2030, "total_steps": 6870, "loss": 0.3286, "lr": 4.440984936465618e-05, "epoch": 0.8864628820960698, "percentage": 29.55, "elapsed_time": "8:04:07", "remaining_time": "19:14:17"} +{"current_steps": 2035, "total_steps": 6870, "loss": 0.3172, "lr": 4.436975801969013e-05, "epoch": 0.888646288209607, "percentage": 29.62, "elapsed_time": "8:05:19", "remaining_time": "19:13:05"} +{"current_steps": 2040, "total_steps": 6870, "loss": 0.3277, "lr": 4.4329541658832454e-05, "epoch": 0.8908296943231441, "percentage": 29.69, "elapsed_time": "8:06:30", "remaining_time": "19:11:53"} +{"current_steps": 2045, "total_steps": 6870, "loss": 0.3371, "lr": 4.428920054164673e-05, "epoch": 0.8930131004366813, "percentage": 29.77, "elapsed_time": "8:07:42", "remaining_time": "19:10:42"} +{"current_steps": 2050, "total_steps": 6870, "loss": 0.3335, "lr": 4.424873492850177e-05, "epoch": 0.8951965065502183, "percentage": 29.84, "elapsed_time": "8:08:54", "remaining_time": "19:09:30"} +{"current_steps": 2055, "total_steps": 6870, "loss": 0.3418, "lr": 4.42081450805699e-05, "epoch": 0.8973799126637555, "percentage": 29.91, "elapsed_time": "8:10:05", "remaining_time": "19:08:19"} +{"current_steps": 2060, "total_steps": 6870, "loss": 0.3215, "lr": 4.4167431259825274e-05, "epoch": 0.8995633187772926, "percentage": 29.99, "elapsed_time": "8:11:17", "remaining_time": "19:07:07"} +{"current_steps": 2065, "total_steps": 6870, "loss": 0.3379, "lr": 4.41265937290422e-05, "epoch": 0.9017467248908297, "percentage": 30.06, "elapsed_time": "8:12:28", "remaining_time": "19:05:56"} +{"current_steps": 2070, "total_steps": 6870, "loss": 0.3169, "lr": 4.408563275179341e-05, "epoch": 0.9039301310043668, "percentage": 30.13, "elapsed_time": "8:13:40", "remaining_time": "19:04:44"} +{"current_steps": 2075, "total_steps": 6870, "loss": 0.3287, "lr": 4.404454859244841e-05, "epoch": 0.9061135371179039, "percentage": 30.2, "elapsed_time": "8:14:51", "remaining_time": "19:03:33"} +{"current_steps": 2080, "total_steps": 6870, "loss": 0.3485, "lr": 4.400334151617171e-05, "epoch": 0.9082969432314411, "percentage": 30.28, "elapsed_time": "8:16:03", "remaining_time": "19:02:21"} +{"current_steps": 2085, "total_steps": 6870, "loss": 0.316, "lr": 4.39620117889212e-05, "epoch": 0.9104803493449781, "percentage": 30.35, "elapsed_time": "8:17:14", "remaining_time": "19:01:09"} +{"current_steps": 2090, "total_steps": 6870, "loss": 0.3221, "lr": 4.3920559677446335e-05, "epoch": 0.9126637554585153, "percentage": 30.42, "elapsed_time": "8:18:26", "remaining_time": "18:59:58"} +{"current_steps": 2095, "total_steps": 6870, "loss": 0.3319, "lr": 4.387898544928648e-05, "epoch": 0.9148471615720524, "percentage": 30.49, "elapsed_time": "8:19:37", "remaining_time": "18:58:46"} +{"current_steps": 2100, "total_steps": 6870, "loss": 0.3313, "lr": 4.383728937276915e-05, "epoch": 0.9170305676855895, "percentage": 30.57, "elapsed_time": "8:20:49", "remaining_time": "18:57:35"} +{"current_steps": 2105, "total_steps": 6870, "loss": 0.3255, "lr": 4.37954717170083e-05, "epoch": 0.9192139737991266, "percentage": 30.64, "elapsed_time": "8:22:00", "remaining_time": "18:56:23"} +{"current_steps": 2110, "total_steps": 6870, "loss": 0.3233, "lr": 4.375353275190259e-05, "epoch": 0.9213973799126638, "percentage": 30.71, "elapsed_time": "8:23:12", "remaining_time": "18:55:12"} +{"current_steps": 2115, "total_steps": 6870, "loss": 0.3341, "lr": 4.371147274813362e-05, "epoch": 0.9235807860262009, "percentage": 30.79, "elapsed_time": "8:24:24", "remaining_time": "18:54:00"} +{"current_steps": 2120, "total_steps": 6870, "loss": 0.3271, "lr": 4.3669291977164185e-05, "epoch": 0.925764192139738, "percentage": 30.86, "elapsed_time": "8:25:35", "remaining_time": "18:52:48"} +{"current_steps": 2125, "total_steps": 6870, "loss": 0.3352, "lr": 4.362699071123655e-05, "epoch": 0.9279475982532751, "percentage": 30.93, "elapsed_time": "8:26:47", "remaining_time": "18:51:37"} +{"current_steps": 2130, "total_steps": 6870, "loss": 0.3221, "lr": 4.358456922337068e-05, "epoch": 0.9301310043668122, "percentage": 31.0, "elapsed_time": "8:27:58", "remaining_time": "18:50:25"} +{"current_steps": 2135, "total_steps": 6870, "loss": 0.3298, "lr": 4.354202778736245e-05, "epoch": 0.9323144104803494, "percentage": 31.08, "elapsed_time": "8:29:10", "remaining_time": "18:49:14"} +{"current_steps": 2140, "total_steps": 6870, "loss": 0.3355, "lr": 4.3499366677781927e-05, "epoch": 0.9344978165938864, "percentage": 31.15, "elapsed_time": "8:30:21", "remaining_time": "18:48:02"} +{"current_steps": 2145, "total_steps": 6870, "loss": 0.3137, "lr": 4.3456586169971555e-05, "epoch": 0.9366812227074236, "percentage": 31.22, "elapsed_time": "8:31:33", "remaining_time": "18:46:51"} +{"current_steps": 2150, "total_steps": 6870, "loss": 0.3392, "lr": 4.341368654004442e-05, "epoch": 0.9388646288209607, "percentage": 31.3, "elapsed_time": "8:32:44", "remaining_time": "18:45:39"} +{"current_steps": 2155, "total_steps": 6870, "loss": 0.3209, "lr": 4.3370668064882406e-05, "epoch": 0.9410480349344978, "percentage": 31.37, "elapsed_time": "8:33:56", "remaining_time": "18:44:28"} +{"current_steps": 2160, "total_steps": 6870, "loss": 0.347, "lr": 4.332753102213447e-05, "epoch": 0.9432314410480349, "percentage": 31.44, "elapsed_time": "8:35:07", "remaining_time": "18:43:16"} +{"current_steps": 2165, "total_steps": 6870, "loss": 0.3223, "lr": 4.328427569021481e-05, "epoch": 0.9454148471615721, "percentage": 31.51, "elapsed_time": "8:36:19", "remaining_time": "18:42:04"} +{"current_steps": 2170, "total_steps": 6870, "loss": 0.3131, "lr": 4.3240902348301134e-05, "epoch": 0.9475982532751092, "percentage": 31.59, "elapsed_time": "8:37:30", "remaining_time": "18:40:53"} +{"current_steps": 2175, "total_steps": 6870, "loss": 0.3254, "lr": 4.319741127633273e-05, "epoch": 0.9497816593886463, "percentage": 31.66, "elapsed_time": "8:38:42", "remaining_time": "18:39:41"} +{"current_steps": 2180, "total_steps": 6870, "loss": 0.3334, "lr": 4.315380275500879e-05, "epoch": 0.9519650655021834, "percentage": 31.73, "elapsed_time": "8:39:54", "remaining_time": "18:38:30"} +{"current_steps": 2185, "total_steps": 6870, "loss": 0.3194, "lr": 4.311007706578654e-05, "epoch": 0.9541484716157205, "percentage": 31.8, "elapsed_time": "8:41:05", "remaining_time": "18:37:18"} +{"current_steps": 2190, "total_steps": 6870, "loss": 0.3171, "lr": 4.306623449087939e-05, "epoch": 0.9563318777292577, "percentage": 31.88, "elapsed_time": "8:42:17", "remaining_time": "18:36:06"} +{"current_steps": 2195, "total_steps": 6870, "loss": 0.3196, "lr": 4.3022275313255215e-05, "epoch": 0.9585152838427947, "percentage": 31.95, "elapsed_time": "8:43:28", "remaining_time": "18:34:55"} +{"current_steps": 2200, "total_steps": 6870, "loss": 0.3128, "lr": 4.2978199816634404e-05, "epoch": 0.9606986899563319, "percentage": 32.02, "elapsed_time": "8:44:40", "remaining_time": "18:33:43"} +{"current_steps": 2205, "total_steps": 6870, "loss": 0.3275, "lr": 4.293400828548812e-05, "epoch": 0.962882096069869, "percentage": 32.1, "elapsed_time": "8:45:51", "remaining_time": "18:32:31"} +{"current_steps": 2210, "total_steps": 6870, "loss": 0.3252, "lr": 4.2889701005036444e-05, "epoch": 0.9650655021834061, "percentage": 32.17, "elapsed_time": "8:47:03", "remaining_time": "18:31:20"} +{"current_steps": 2215, "total_steps": 6870, "loss": 0.3394, "lr": 4.28452782612465e-05, "epoch": 0.9672489082969432, "percentage": 32.24, "elapsed_time": "8:48:14", "remaining_time": "18:30:08"} +{"current_steps": 2220, "total_steps": 6870, "loss": 0.3127, "lr": 4.2800740340830636e-05, "epoch": 0.9694323144104804, "percentage": 32.31, "elapsed_time": "8:49:26", "remaining_time": "18:28:57"} +{"current_steps": 2225, "total_steps": 6870, "loss": 0.3131, "lr": 4.275608753124459e-05, "epoch": 0.9716157205240175, "percentage": 32.39, "elapsed_time": "8:50:37", "remaining_time": "18:27:45"} +{"current_steps": 2230, "total_steps": 6870, "loss": 0.314, "lr": 4.271132012068559e-05, "epoch": 0.9737991266375546, "percentage": 32.46, "elapsed_time": "8:51:49", "remaining_time": "18:26:34"} +{"current_steps": 2235, "total_steps": 6870, "loss": 0.3213, "lr": 4.266643839809054e-05, "epoch": 0.9759825327510917, "percentage": 32.53, "elapsed_time": "8:53:00", "remaining_time": "18:25:22"} +{"current_steps": 2240, "total_steps": 6870, "loss": 0.326, "lr": 4.262144265313412e-05, "epoch": 0.9781659388646288, "percentage": 32.61, "elapsed_time": "8:54:12", "remaining_time": "18:24:10"} +{"current_steps": 2245, "total_steps": 6870, "loss": 0.3168, "lr": 4.2576333176226944e-05, "epoch": 0.980349344978166, "percentage": 32.68, "elapsed_time": "8:55:23", "remaining_time": "18:22:59"} +{"current_steps": 2250, "total_steps": 6870, "loss": 0.3317, "lr": 4.253111025851366e-05, "epoch": 0.982532751091703, "percentage": 32.75, "elapsed_time": "8:56:35", "remaining_time": "18:21:47"} +{"current_steps": 2255, "total_steps": 6870, "loss": 0.3126, "lr": 4.2485774191871084e-05, "epoch": 0.9847161572052402, "percentage": 32.82, "elapsed_time": "8:57:46", "remaining_time": "18:20:35"} +{"current_steps": 2260, "total_steps": 6870, "loss": 0.3181, "lr": 4.244032526890633e-05, "epoch": 0.9868995633187773, "percentage": 32.9, "elapsed_time": "8:58:58", "remaining_time": "18:19:24"} +{"current_steps": 2265, "total_steps": 6870, "loss": 0.3186, "lr": 4.239476378295489e-05, "epoch": 0.9890829694323144, "percentage": 32.97, "elapsed_time": "9:00:09", "remaining_time": "18:18:12"} +{"current_steps": 2270, "total_steps": 6870, "loss": 0.3093, "lr": 4.2349090028078766e-05, "epoch": 0.9912663755458515, "percentage": 33.04, "elapsed_time": "9:01:21", "remaining_time": "18:17:01"} +{"current_steps": 2275, "total_steps": 6870, "loss": 0.3082, "lr": 4.230330429906456e-05, "epoch": 0.9934497816593887, "percentage": 33.11, "elapsed_time": "9:02:32", "remaining_time": "18:15:49"} +{"current_steps": 2280, "total_steps": 6870, "loss": 0.308, "lr": 4.225740689142158e-05, "epoch": 0.9956331877729258, "percentage": 33.19, "elapsed_time": "9:03:44", "remaining_time": "18:14:38"} +{"current_steps": 2285, "total_steps": 6870, "loss": 0.3125, "lr": 4.221139810137992e-05, "epoch": 0.9978165938864629, "percentage": 33.26, "elapsed_time": "9:04:55", "remaining_time": "18:13:26"} +{"current_steps": 2290, "total_steps": 6870, "loss": 0.3134, "lr": 4.216527822588857e-05, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "9:06:07", "remaining_time": "18:12:14"} +{"current_steps": 2295, "total_steps": 6870, "loss": 0.2917, "lr": 4.2119047562613476e-05, "epoch": 1.002183406113537, "percentage": 33.41, "elapsed_time": "9:07:18", "remaining_time": "18:11:03"} +{"current_steps": 2300, "total_steps": 6870, "loss": 0.2763, "lr": 4.207270640993561e-05, "epoch": 1.0043668122270741, "percentage": 33.48, "elapsed_time": "9:08:30", "remaining_time": "18:09:51"} +{"current_steps": 2305, "total_steps": 6870, "loss": 0.267, "lr": 4.202625506694911e-05, "epoch": 1.0065502183406114, "percentage": 33.55, "elapsed_time": "9:09:41", "remaining_time": "18:08:40"} +{"current_steps": 2310, "total_steps": 6870, "loss": 0.2785, "lr": 4.197969383345925e-05, "epoch": 1.0087336244541485, "percentage": 33.62, "elapsed_time": "9:10:53", "remaining_time": "18:07:28"} +{"current_steps": 2315, "total_steps": 6870, "loss": 0.2961, "lr": 4.19330230099806e-05, "epoch": 1.0109170305676856, "percentage": 33.7, "elapsed_time": "9:12:05", "remaining_time": "18:06:16"} +{"current_steps": 2320, "total_steps": 6870, "loss": 0.2949, "lr": 4.188624289773501e-05, "epoch": 1.0131004366812226, "percentage": 33.77, "elapsed_time": "9:13:16", "remaining_time": "18:05:05"} +{"current_steps": 2325, "total_steps": 6870, "loss": 0.2738, "lr": 4.1839353798649717e-05, "epoch": 1.01528384279476, "percentage": 33.84, "elapsed_time": "9:14:28", "remaining_time": "18:03:53"} +{"current_steps": 2330, "total_steps": 6870, "loss": 0.2755, "lr": 4.1792356015355375e-05, "epoch": 1.017467248908297, "percentage": 33.92, "elapsed_time": "9:15:39", "remaining_time": "18:02:42"} +{"current_steps": 2335, "total_steps": 6870, "loss": 0.2845, "lr": 4.17452498511841e-05, "epoch": 1.019650655021834, "percentage": 33.99, "elapsed_time": "9:16:51", "remaining_time": "18:01:30"} +{"current_steps": 2340, "total_steps": 6870, "loss": 0.2808, "lr": 4.169803561016754e-05, "epoch": 1.0218340611353711, "percentage": 34.06, "elapsed_time": "9:18:02", "remaining_time": "18:00:19"} +{"current_steps": 2345, "total_steps": 6870, "loss": 0.2849, "lr": 4.1650713597034845e-05, "epoch": 1.0240174672489082, "percentage": 34.13, "elapsed_time": "9:19:14", "remaining_time": "17:59:07"} +{"current_steps": 2350, "total_steps": 6870, "loss": 0.2914, "lr": 4.16032841172108e-05, "epoch": 1.0262008733624455, "percentage": 34.21, "elapsed_time": "9:20:25", "remaining_time": "17:57:56"} +{"current_steps": 2355, "total_steps": 6870, "loss": 0.2847, "lr": 4.155574747681375e-05, "epoch": 1.0283842794759825, "percentage": 34.28, "elapsed_time": "9:21:37", "remaining_time": "17:56:44"} +{"current_steps": 2360, "total_steps": 6870, "loss": 0.2984, "lr": 4.1508103982653715e-05, "epoch": 1.0305676855895196, "percentage": 34.35, "elapsed_time": "9:22:48", "remaining_time": "17:55:33"} +{"current_steps": 2365, "total_steps": 6870, "loss": 0.2761, "lr": 4.146035394223034e-05, "epoch": 1.0327510917030567, "percentage": 34.43, "elapsed_time": "9:24:00", "remaining_time": "17:54:21"} +{"current_steps": 2370, "total_steps": 6870, "loss": 0.2737, "lr": 4.1412497663730954e-05, "epoch": 1.034934497816594, "percentage": 34.5, "elapsed_time": "9:25:12", "remaining_time": "17:53:10"} +{"current_steps": 2375, "total_steps": 6870, "loss": 0.2837, "lr": 4.1364535456028555e-05, "epoch": 1.037117903930131, "percentage": 34.57, "elapsed_time": "9:26:23", "remaining_time": "17:51:58"} +{"current_steps": 2380, "total_steps": 6870, "loss": 0.2613, "lr": 4.1316467628679834e-05, "epoch": 1.039301310043668, "percentage": 34.64, "elapsed_time": "9:27:35", "remaining_time": "17:50:46"} +{"current_steps": 2385, "total_steps": 6870, "loss": 0.2846, "lr": 4.126829449192316e-05, "epoch": 1.0414847161572052, "percentage": 34.72, "elapsed_time": "9:28:46", "remaining_time": "17:49:35"} +{"current_steps": 2390, "total_steps": 6870, "loss": 0.2783, "lr": 4.1220016356676615e-05, "epoch": 1.0436681222707425, "percentage": 34.79, "elapsed_time": "9:29:58", "remaining_time": "17:48:23"} +{"current_steps": 2395, "total_steps": 6870, "loss": 0.2822, "lr": 4.117163353453593e-05, "epoch": 1.0458515283842795, "percentage": 34.86, "elapsed_time": "9:31:09", "remaining_time": "17:47:12"} +{"current_steps": 2400, "total_steps": 6870, "loss": 0.2831, "lr": 4.112314633777251e-05, "epoch": 1.0480349344978166, "percentage": 34.93, "elapsed_time": "9:32:21", "remaining_time": "17:46:00"} +{"current_steps": 2405, "total_steps": 6870, "loss": 0.2735, "lr": 4.107455507933144e-05, "epoch": 1.0502183406113537, "percentage": 35.01, "elapsed_time": "9:33:32", "remaining_time": "17:44:49"} +{"current_steps": 2410, "total_steps": 6870, "loss": 0.2784, "lr": 4.102586007282938e-05, "epoch": 1.0524017467248907, "percentage": 35.08, "elapsed_time": "9:34:44", "remaining_time": "17:43:37"} +{"current_steps": 2415, "total_steps": 6870, "loss": 0.2815, "lr": 4.0977061632552656e-05, "epoch": 1.054585152838428, "percentage": 35.15, "elapsed_time": "9:35:56", "remaining_time": "17:42:26"} +{"current_steps": 2420, "total_steps": 6870, "loss": 0.291, "lr": 4.092816007345515e-05, "epoch": 1.056768558951965, "percentage": 35.23, "elapsed_time": "9:37:07", "remaining_time": "17:41:14"} +{"current_steps": 2425, "total_steps": 6870, "loss": 0.2653, "lr": 4.087915571115629e-05, "epoch": 1.0589519650655022, "percentage": 35.3, "elapsed_time": "9:38:19", "remaining_time": "17:40:03"} +{"current_steps": 2430, "total_steps": 6870, "loss": 0.2816, "lr": 4.083004886193902e-05, "epoch": 1.0611353711790392, "percentage": 35.37, "elapsed_time": "9:39:30", "remaining_time": "17:38:51"} +{"current_steps": 2435, "total_steps": 6870, "loss": 0.2871, "lr": 4.0780839842747754e-05, "epoch": 1.0633187772925765, "percentage": 35.44, "elapsed_time": "9:40:42", "remaining_time": "17:37:40"} +{"current_steps": 2440, "total_steps": 6870, "loss": 0.2901, "lr": 4.073152897118632e-05, "epoch": 1.0655021834061136, "percentage": 35.52, "elapsed_time": "9:41:53", "remaining_time": "17:36:28"} +{"current_steps": 2445, "total_steps": 6870, "loss": 0.2937, "lr": 4.0682116565515935e-05, "epoch": 1.0676855895196506, "percentage": 35.59, "elapsed_time": "9:43:05", "remaining_time": "17:35:16"} +{"current_steps": 2450, "total_steps": 6870, "loss": 0.2745, "lr": 4.063260294465311e-05, "epoch": 1.0698689956331877, "percentage": 35.66, "elapsed_time": "9:44:16", "remaining_time": "17:34:05"} +{"current_steps": 2455, "total_steps": 6870, "loss": 0.2751, "lr": 4.058298842816765e-05, "epoch": 1.072052401746725, "percentage": 35.74, "elapsed_time": "9:45:28", "remaining_time": "17:32:53"} +{"current_steps": 2460, "total_steps": 6870, "loss": 0.2745, "lr": 4.0533273336280516e-05, "epoch": 1.074235807860262, "percentage": 35.81, "elapsed_time": "9:46:39", "remaining_time": "17:31:42"} +{"current_steps": 2465, "total_steps": 6870, "loss": 0.2929, "lr": 4.048345798986186e-05, "epoch": 1.0764192139737991, "percentage": 35.88, "elapsed_time": "9:47:51", "remaining_time": "17:30:30"} +{"current_steps": 2470, "total_steps": 6870, "loss": 0.2771, "lr": 4.043354271042884e-05, "epoch": 1.0786026200873362, "percentage": 35.95, "elapsed_time": "9:49:03", "remaining_time": "17:29:19"} +{"current_steps": 2475, "total_steps": 6870, "loss": 0.2778, "lr": 4.038352782014362e-05, "epoch": 1.0807860262008733, "percentage": 36.03, "elapsed_time": "9:50:14", "remaining_time": "17:28:07"} +{"current_steps": 2480, "total_steps": 6870, "loss": 0.2843, "lr": 4.033341364181128e-05, "epoch": 1.0829694323144106, "percentage": 36.1, "elapsed_time": "9:51:26", "remaining_time": "17:26:56"} +{"current_steps": 2485, "total_steps": 6870, "loss": 0.281, "lr": 4.028320049887771e-05, "epoch": 1.0851528384279476, "percentage": 36.17, "elapsed_time": "9:52:37", "remaining_time": "17:25:44"} +{"current_steps": 2490, "total_steps": 6870, "loss": 0.2762, "lr": 4.023288871542752e-05, "epoch": 1.0873362445414847, "percentage": 36.24, "elapsed_time": "9:53:49", "remaining_time": "17:24:33"} +{"current_steps": 2495, "total_steps": 6870, "loss": 0.2803, "lr": 4.0182478616182015e-05, "epoch": 1.0895196506550218, "percentage": 36.32, "elapsed_time": "9:55:00", "remaining_time": "17:23:21"} +{"current_steps": 2500, "total_steps": 6870, "loss": 0.2882, "lr": 4.013197052649699e-05, "epoch": 1.091703056768559, "percentage": 36.39, "elapsed_time": "9:56:12", "remaining_time": "17:22:10"} +{"current_steps": 2505, "total_steps": 6870, "loss": 0.2845, "lr": 4.0081364772360725e-05, "epoch": 1.0938864628820961, "percentage": 36.46, "elapsed_time": "9:57:23", "remaining_time": "17:20:58"} +{"current_steps": 2510, "total_steps": 6870, "loss": 0.2742, "lr": 4.003066168039183e-05, "epoch": 1.0960698689956332, "percentage": 36.54, "elapsed_time": "9:58:35", "remaining_time": "17:19:46"} +{"current_steps": 2515, "total_steps": 6870, "loss": 0.2777, "lr": 3.9979861577837155e-05, "epoch": 1.0982532751091703, "percentage": 36.61, "elapsed_time": "9:59:46", "remaining_time": "17:18:35"} +{"current_steps": 2520, "total_steps": 6870, "loss": 0.2809, "lr": 3.9928964792569655e-05, "epoch": 1.1004366812227073, "percentage": 36.68, "elapsed_time": "10:00:58", "remaining_time": "17:17:23"} +{"current_steps": 2525, "total_steps": 6870, "loss": 0.2799, "lr": 3.987797165308633e-05, "epoch": 1.1026200873362446, "percentage": 36.75, "elapsed_time": "10:02:10", "remaining_time": "17:16:12"} +{"current_steps": 2530, "total_steps": 6870, "loss": 0.2924, "lr": 3.982688248850604e-05, "epoch": 1.1048034934497817, "percentage": 36.83, "elapsed_time": "10:03:21", "remaining_time": "17:15:00"} +{"current_steps": 2535, "total_steps": 6870, "loss": 0.29, "lr": 3.977569762856741e-05, "epoch": 1.1069868995633187, "percentage": 36.9, "elapsed_time": "10:04:33", "remaining_time": "17:13:49"} +{"current_steps": 2540, "total_steps": 6870, "loss": 0.283, "lr": 3.9724417403626705e-05, "epoch": 1.1091703056768558, "percentage": 36.97, "elapsed_time": "10:05:44", "remaining_time": "17:12:38"} +{"current_steps": 2545, "total_steps": 6870, "loss": 0.2928, "lr": 3.9673042144655697e-05, "epoch": 1.111353711790393, "percentage": 37.05, "elapsed_time": "10:06:56", "remaining_time": "17:11:26"} +{"current_steps": 2550, "total_steps": 6870, "loss": 0.2775, "lr": 3.9621572183239515e-05, "epoch": 1.1135371179039302, "percentage": 37.12, "elapsed_time": "10:08:07", "remaining_time": "17:10:14"} +{"current_steps": 2555, "total_steps": 6870, "loss": 0.3014, "lr": 3.957000785157452e-05, "epoch": 1.1157205240174672, "percentage": 37.19, "elapsed_time": "10:09:19", "remaining_time": "17:09:03"} +{"current_steps": 2560, "total_steps": 6870, "loss": 0.268, "lr": 3.951834948246616e-05, "epoch": 1.1179039301310043, "percentage": 37.26, "elapsed_time": "10:10:31", "remaining_time": "17:07:51"} +{"current_steps": 2565, "total_steps": 6870, "loss": 0.2997, "lr": 3.946659740932681e-05, "epoch": 1.1200873362445414, "percentage": 37.34, "elapsed_time": "10:11:42", "remaining_time": "17:06:40"} +{"current_steps": 2570, "total_steps": 6870, "loss": 0.2785, "lr": 3.9414751966173634e-05, "epoch": 1.1222707423580787, "percentage": 37.41, "elapsed_time": "10:12:54", "remaining_time": "17:05:28"} +{"current_steps": 2575, "total_steps": 6870, "loss": 0.2759, "lr": 3.93628134876264e-05, "epoch": 1.1244541484716157, "percentage": 37.48, "elapsed_time": "10:14:05", "remaining_time": "17:04:17"} +{"current_steps": 2580, "total_steps": 6870, "loss": 0.2804, "lr": 3.9310782308905395e-05, "epoch": 1.1266375545851528, "percentage": 37.55, "elapsed_time": "10:15:17", "remaining_time": "17:03:05"} +{"current_steps": 2585, "total_steps": 6870, "loss": 0.2669, "lr": 3.925865876582915e-05, "epoch": 1.1288209606986899, "percentage": 37.63, "elapsed_time": "10:16:28", "remaining_time": "17:01:54"} +{"current_steps": 2590, "total_steps": 6870, "loss": 0.2756, "lr": 3.920644319481237e-05, "epoch": 1.1310043668122272, "percentage": 37.7, "elapsed_time": "10:17:40", "remaining_time": "17:00:42"} +{"current_steps": 2595, "total_steps": 6870, "loss": 0.283, "lr": 3.915413593286371e-05, "epoch": 1.1331877729257642, "percentage": 37.77, "elapsed_time": "10:18:51", "remaining_time": "16:59:31"} +{"current_steps": 2600, "total_steps": 6870, "loss": 0.2749, "lr": 3.910173731758362e-05, "epoch": 1.1353711790393013, "percentage": 37.85, "elapsed_time": "10:20:03", "remaining_time": "16:58:19"} +{"current_steps": 2605, "total_steps": 6870, "loss": 0.2759, "lr": 3.904924768716216e-05, "epoch": 1.1375545851528384, "percentage": 37.92, "elapsed_time": "10:21:15", "remaining_time": "16:57:08"} +{"current_steps": 2610, "total_steps": 6870, "loss": 0.2745, "lr": 3.89966673803768e-05, "epoch": 1.1397379912663754, "percentage": 37.99, "elapsed_time": "10:22:26", "remaining_time": "16:55:56"} +{"current_steps": 2615, "total_steps": 6870, "loss": 0.2903, "lr": 3.894399673659031e-05, "epoch": 1.1419213973799127, "percentage": 38.06, "elapsed_time": "10:23:38", "remaining_time": "16:54:45"} +{"current_steps": 2620, "total_steps": 6870, "loss": 0.2687, "lr": 3.889123609574843e-05, "epoch": 1.1441048034934498, "percentage": 38.14, "elapsed_time": "10:24:49", "remaining_time": "16:53:33"} +{"current_steps": 2625, "total_steps": 6870, "loss": 0.2893, "lr": 3.883838579837781e-05, "epoch": 1.1462882096069869, "percentage": 38.21, "elapsed_time": "10:26:01", "remaining_time": "16:52:22"} +{"current_steps": 2630, "total_steps": 6870, "loss": 0.2848, "lr": 3.8785446185583755e-05, "epoch": 1.1484716157205241, "percentage": 38.28, "elapsed_time": "10:27:12", "remaining_time": "16:51:10"} +{"current_steps": 2635, "total_steps": 6870, "loss": 0.2845, "lr": 3.873241759904801e-05, "epoch": 1.1506550218340612, "percentage": 38.36, "elapsed_time": "10:28:24", "remaining_time": "16:49:59"} +{"current_steps": 2640, "total_steps": 6870, "loss": 0.2926, "lr": 3.8679300381026585e-05, "epoch": 1.1528384279475983, "percentage": 38.43, "elapsed_time": "10:29:36", "remaining_time": "16:48:47"} +{"current_steps": 2645, "total_steps": 6870, "loss": 0.2698, "lr": 3.862609487434752e-05, "epoch": 1.1550218340611353, "percentage": 38.5, "elapsed_time": "10:30:47", "remaining_time": "16:47:36"} +{"current_steps": 2650, "total_steps": 6870, "loss": 0.2814, "lr": 3.857280142240871e-05, "epoch": 1.1572052401746724, "percentage": 38.57, "elapsed_time": "10:31:59", "remaining_time": "16:46:24"} +{"current_steps": 2655, "total_steps": 6870, "loss": 0.2988, "lr": 3.851942036917565e-05, "epoch": 1.1593886462882097, "percentage": 38.65, "elapsed_time": "10:33:10", "remaining_time": "16:45:12"} +{"current_steps": 2660, "total_steps": 6870, "loss": 0.2828, "lr": 3.846595205917922e-05, "epoch": 1.1615720524017468, "percentage": 38.72, "elapsed_time": "10:34:22", "remaining_time": "16:44:01"} +{"current_steps": 2665, "total_steps": 6870, "loss": 0.284, "lr": 3.8412396837513484e-05, "epoch": 1.1637554585152838, "percentage": 38.79, "elapsed_time": "10:35:33", "remaining_time": "16:42:49"} +{"current_steps": 2670, "total_steps": 6870, "loss": 0.2895, "lr": 3.835875504983344e-05, "epoch": 1.165938864628821, "percentage": 38.86, "elapsed_time": "10:36:45", "remaining_time": "16:41:38"} +{"current_steps": 2675, "total_steps": 6870, "loss": 0.281, "lr": 3.830502704235282e-05, "epoch": 1.1681222707423582, "percentage": 38.94, "elapsed_time": "10:37:57", "remaining_time": "16:40:26"} +{"current_steps": 2680, "total_steps": 6870, "loss": 0.2889, "lr": 3.825121316184181e-05, "epoch": 1.1703056768558953, "percentage": 39.01, "elapsed_time": "10:39:08", "remaining_time": "16:39:15"} +{"current_steps": 2685, "total_steps": 6870, "loss": 0.266, "lr": 3.8197313755624855e-05, "epoch": 1.1724890829694323, "percentage": 39.08, "elapsed_time": "10:40:20", "remaining_time": "16:38:03"} +{"current_steps": 2690, "total_steps": 6870, "loss": 0.2791, "lr": 3.8143329171578366e-05, "epoch": 1.1746724890829694, "percentage": 39.16, "elapsed_time": "10:41:31", "remaining_time": "16:36:52"} +{"current_steps": 2695, "total_steps": 6870, "loss": 0.2972, "lr": 3.808925975812855e-05, "epoch": 1.1768558951965065, "percentage": 39.23, "elapsed_time": "10:42:43", "remaining_time": "16:35:40"} +{"current_steps": 2700, "total_steps": 6870, "loss": 0.2902, "lr": 3.803510586424908e-05, "epoch": 1.1790393013100438, "percentage": 39.3, "elapsed_time": "10:43:54", "remaining_time": "16:34:29"} +{"current_steps": 2705, "total_steps": 6870, "loss": 0.282, "lr": 3.798086783945892e-05, "epoch": 1.1812227074235808, "percentage": 39.37, "elapsed_time": "10:45:06", "remaining_time": "16:33:17"} +{"current_steps": 2710, "total_steps": 6870, "loss": 0.2837, "lr": 3.792654603381998e-05, "epoch": 1.1834061135371179, "percentage": 39.45, "elapsed_time": "10:46:17", "remaining_time": "16:32:06"} +{"current_steps": 2715, "total_steps": 6870, "loss": 0.2712, "lr": 3.7872140797934956e-05, "epoch": 1.185589519650655, "percentage": 39.52, "elapsed_time": "10:47:29", "remaining_time": "16:30:54"} +{"current_steps": 2720, "total_steps": 6870, "loss": 0.2754, "lr": 3.781765248294499e-05, "epoch": 1.1877729257641922, "percentage": 39.59, "elapsed_time": "10:48:41", "remaining_time": "16:29:43"} +{"current_steps": 2725, "total_steps": 6870, "loss": 0.2854, "lr": 3.7763081440527435e-05, "epoch": 1.1899563318777293, "percentage": 39.67, "elapsed_time": "10:49:52", "remaining_time": "16:28:31"} +{"current_steps": 2730, "total_steps": 6870, "loss": 0.2858, "lr": 3.7708428022893596e-05, "epoch": 1.1921397379912664, "percentage": 39.74, "elapsed_time": "10:51:04", "remaining_time": "16:27:20"} +{"current_steps": 2735, "total_steps": 6870, "loss": 0.2718, "lr": 3.7653692582786416e-05, "epoch": 1.1943231441048034, "percentage": 39.81, "elapsed_time": "10:52:15", "remaining_time": "16:26:08"} +{"current_steps": 2740, "total_steps": 6870, "loss": 0.2717, "lr": 3.7598875473478246e-05, "epoch": 1.1965065502183405, "percentage": 39.88, "elapsed_time": "10:53:27", "remaining_time": "16:24:57"} +{"current_steps": 2745, "total_steps": 6870, "loss": 0.271, "lr": 3.7543977048768534e-05, "epoch": 1.1986899563318778, "percentage": 39.96, "elapsed_time": "10:54:38", "remaining_time": "16:23:45"} +{"current_steps": 2750, "total_steps": 6870, "loss": 0.2847, "lr": 3.748899766298157e-05, "epoch": 1.2008733624454149, "percentage": 40.03, "elapsed_time": "10:55:50", "remaining_time": "16:22:33"} +{"current_steps": 2755, "total_steps": 6870, "loss": 0.2754, "lr": 3.743393767096416e-05, "epoch": 1.203056768558952, "percentage": 40.1, "elapsed_time": "10:57:01", "remaining_time": "16:21:22"} +{"current_steps": 2760, "total_steps": 6870, "loss": 0.2788, "lr": 3.737879742808337e-05, "epoch": 1.205240174672489, "percentage": 40.17, "elapsed_time": "10:58:13", "remaining_time": "16:20:10"} +{"current_steps": 2765, "total_steps": 6870, "loss": 0.2872, "lr": 3.732357729022421e-05, "epoch": 1.2074235807860263, "percentage": 40.25, "elapsed_time": "10:59:25", "remaining_time": "16:18:59"} +{"current_steps": 2770, "total_steps": 6870, "loss": 0.2869, "lr": 3.7268277613787356e-05, "epoch": 1.2096069868995634, "percentage": 40.32, "elapsed_time": "11:00:36", "remaining_time": "16:17:47"} +{"current_steps": 2775, "total_steps": 6870, "loss": 0.2852, "lr": 3.721289875568683e-05, "epoch": 1.2117903930131004, "percentage": 40.39, "elapsed_time": "11:01:48", "remaining_time": "16:16:36"} +{"current_steps": 2780, "total_steps": 6870, "loss": 0.27, "lr": 3.715744107334773e-05, "epoch": 1.2139737991266375, "percentage": 40.47, "elapsed_time": "11:02:59", "remaining_time": "16:15:24"} +{"current_steps": 2785, "total_steps": 6870, "loss": 0.2823, "lr": 3.7101904924703865e-05, "epoch": 1.2161572052401746, "percentage": 40.54, "elapsed_time": "11:04:11", "remaining_time": "16:14:13"} +{"current_steps": 2790, "total_steps": 6870, "loss": 0.2771, "lr": 3.704629066819549e-05, "epoch": 1.2183406113537119, "percentage": 40.61, "elapsed_time": "11:05:22", "remaining_time": "16:13:01"} +{"current_steps": 2795, "total_steps": 6870, "loss": 0.2698, "lr": 3.6990598662767014e-05, "epoch": 1.220524017467249, "percentage": 40.68, "elapsed_time": "11:06:34", "remaining_time": "16:11:50"} +{"current_steps": 2800, "total_steps": 6870, "loss": 0.2725, "lr": 3.693482926786461e-05, "epoch": 1.222707423580786, "percentage": 40.76, "elapsed_time": "11:07:45", "remaining_time": "16:10:38"} +{"current_steps": 2805, "total_steps": 6870, "loss": 0.2892, "lr": 3.687898284343396e-05, "epoch": 1.224890829694323, "percentage": 40.83, "elapsed_time": "11:08:57", "remaining_time": "16:09:26"} +{"current_steps": 2810, "total_steps": 6870, "loss": 0.2804, "lr": 3.68230597499179e-05, "epoch": 1.2270742358078603, "percentage": 40.9, "elapsed_time": "11:10:08", "remaining_time": "16:08:15"} +{"current_steps": 2815, "total_steps": 6870, "loss": 0.3014, "lr": 3.676706034825409e-05, "epoch": 1.2292576419213974, "percentage": 40.98, "elapsed_time": "11:11:20", "remaining_time": "16:07:03"} +{"current_steps": 2820, "total_steps": 6870, "loss": 0.2748, "lr": 3.6710984999872735e-05, "epoch": 1.2314410480349345, "percentage": 41.05, "elapsed_time": "11:12:31", "remaining_time": "16:05:52"} +{"current_steps": 2825, "total_steps": 6870, "loss": 0.2903, "lr": 3.665483406669418e-05, "epoch": 1.2336244541484715, "percentage": 41.12, "elapsed_time": "11:13:43", "remaining_time": "16:04:40"} +{"current_steps": 2830, "total_steps": 6870, "loss": 0.2853, "lr": 3.6598607911126614e-05, "epoch": 1.2358078602620086, "percentage": 41.19, "elapsed_time": "11:14:54", "remaining_time": "16:03:28"} +{"current_steps": 2835, "total_steps": 6870, "loss": 0.2661, "lr": 3.654230689606372e-05, "epoch": 1.237991266375546, "percentage": 41.27, "elapsed_time": "11:16:06", "remaining_time": "16:02:17"} +{"current_steps": 2840, "total_steps": 6870, "loss": 0.2776, "lr": 3.648593138488235e-05, "epoch": 1.240174672489083, "percentage": 41.34, "elapsed_time": "11:17:17", "remaining_time": "16:01:05"} +{"current_steps": 2845, "total_steps": 6870, "loss": 0.2807, "lr": 3.6429481741440176e-05, "epoch": 1.24235807860262, "percentage": 41.41, "elapsed_time": "11:18:29", "remaining_time": "15:59:54"} +{"current_steps": 2850, "total_steps": 6870, "loss": 0.2781, "lr": 3.63729583300733e-05, "epoch": 1.244541484716157, "percentage": 41.48, "elapsed_time": "11:19:40", "remaining_time": "15:58:42"} +{"current_steps": 2855, "total_steps": 6870, "loss": 0.2871, "lr": 3.6316361515593966e-05, "epoch": 1.2467248908296944, "percentage": 41.56, "elapsed_time": "11:20:52", "remaining_time": "15:57:30"} +{"current_steps": 2860, "total_steps": 6870, "loss": 0.2761, "lr": 3.625969166328815e-05, "epoch": 1.2489082969432315, "percentage": 41.63, "elapsed_time": "11:22:03", "remaining_time": "15:56:19"} +{"current_steps": 2865, "total_steps": 6870, "loss": 0.289, "lr": 3.620294913891326e-05, "epoch": 1.2510917030567685, "percentage": 41.7, "elapsed_time": "11:23:15", "remaining_time": "15:55:07"} +{"current_steps": 2870, "total_steps": 6870, "loss": 0.2855, "lr": 3.614613430869571e-05, "epoch": 1.2532751091703056, "percentage": 41.78, "elapsed_time": "11:24:27", "remaining_time": "15:53:56"} +{"current_steps": 2875, "total_steps": 6870, "loss": 0.2881, "lr": 3.608924753932862e-05, "epoch": 1.2554585152838427, "percentage": 41.85, "elapsed_time": "11:25:38", "remaining_time": "15:52:45"} +{"current_steps": 2880, "total_steps": 6870, "loss": 0.2859, "lr": 3.6032289197969374e-05, "epoch": 1.25764192139738, "percentage": 41.92, "elapsed_time": "11:26:50", "remaining_time": "15:51:33"} +{"current_steps": 2885, "total_steps": 6870, "loss": 0.2871, "lr": 3.5975259652237333e-05, "epoch": 1.259825327510917, "percentage": 41.99, "elapsed_time": "11:28:01", "remaining_time": "15:50:21"} +{"current_steps": 2890, "total_steps": 6870, "loss": 0.2732, "lr": 3.5918159270211424e-05, "epoch": 1.262008733624454, "percentage": 42.07, "elapsed_time": "11:29:13", "remaining_time": "15:49:10"} +{"current_steps": 2895, "total_steps": 6870, "loss": 0.2772, "lr": 3.586098842042773e-05, "epoch": 1.2641921397379914, "percentage": 42.14, "elapsed_time": "11:30:24", "remaining_time": "15:47:58"} +{"current_steps": 2900, "total_steps": 6870, "loss": 0.2778, "lr": 3.580374747187719e-05, "epoch": 1.2663755458515285, "percentage": 42.21, "elapsed_time": "11:31:36", "remaining_time": "15:46:47"} +{"current_steps": 2905, "total_steps": 6870, "loss": 0.2748, "lr": 3.574643679400313e-05, "epoch": 1.2685589519650655, "percentage": 42.29, "elapsed_time": "11:32:48", "remaining_time": "15:45:35"} +{"current_steps": 2910, "total_steps": 6870, "loss": 0.2883, "lr": 3.568905675669895e-05, "epoch": 1.2707423580786026, "percentage": 42.36, "elapsed_time": "11:33:59", "remaining_time": "15:44:24"} +{"current_steps": 2915, "total_steps": 6870, "loss": 0.2661, "lr": 3.5631607730305694e-05, "epoch": 1.2729257641921397, "percentage": 42.43, "elapsed_time": "11:35:11", "remaining_time": "15:43:12"} +{"current_steps": 2920, "total_steps": 6870, "loss": 0.2797, "lr": 3.5574090085609696e-05, "epoch": 1.2751091703056767, "percentage": 42.5, "elapsed_time": "11:36:22", "remaining_time": "15:42:01"} +{"current_steps": 2925, "total_steps": 6870, "loss": 0.282, "lr": 3.551650419384013e-05, "epoch": 1.277292576419214, "percentage": 42.58, "elapsed_time": "11:37:34", "remaining_time": "15:40:49"} +{"current_steps": 2930, "total_steps": 6870, "loss": 0.2737, "lr": 3.545885042666669e-05, "epoch": 1.279475982532751, "percentage": 42.65, "elapsed_time": "11:38:45", "remaining_time": "15:39:38"} +{"current_steps": 2935, "total_steps": 6870, "loss": 0.2834, "lr": 3.5401129156197105e-05, "epoch": 1.2816593886462881, "percentage": 42.72, "elapsed_time": "11:39:57", "remaining_time": "15:38:26"} +{"current_steps": 2940, "total_steps": 6870, "loss": 0.2818, "lr": 3.534334075497482e-05, "epoch": 1.2838427947598254, "percentage": 42.79, "elapsed_time": "11:41:09", "remaining_time": "15:37:15"} +{"current_steps": 2945, "total_steps": 6870, "loss": 0.282, "lr": 3.5285485595976534e-05, "epoch": 1.2860262008733625, "percentage": 42.87, "elapsed_time": "11:42:20", "remaining_time": "15:36:03"} +{"current_steps": 2950, "total_steps": 6870, "loss": 0.272, "lr": 3.5227564052609816e-05, "epoch": 1.2882096069868996, "percentage": 42.94, "elapsed_time": "11:43:32", "remaining_time": "15:34:51"} +{"current_steps": 2955, "total_steps": 6870, "loss": 0.2785, "lr": 3.5169576498710696e-05, "epoch": 1.2903930131004366, "percentage": 43.01, "elapsed_time": "11:44:43", "remaining_time": "15:33:40"} +{"current_steps": 2960, "total_steps": 6870, "loss": 0.268, "lr": 3.511152330854125e-05, "epoch": 1.2925764192139737, "percentage": 43.09, "elapsed_time": "11:45:55", "remaining_time": "15:32:28"} +{"current_steps": 2965, "total_steps": 6870, "loss": 0.2815, "lr": 3.505340485678717e-05, "epoch": 1.294759825327511, "percentage": 43.16, "elapsed_time": "11:47:06", "remaining_time": "15:31:17"} +{"current_steps": 2970, "total_steps": 6870, "loss": 0.2978, "lr": 3.499522151855537e-05, "epoch": 1.296943231441048, "percentage": 43.23, "elapsed_time": "11:48:18", "remaining_time": "15:30:05"} +{"current_steps": 2975, "total_steps": 6870, "loss": 0.2695, "lr": 3.493697366937154e-05, "epoch": 1.2991266375545851, "percentage": 43.3, "elapsed_time": "11:49:29", "remaining_time": "15:28:54"} +{"current_steps": 2980, "total_steps": 6870, "loss": 0.2938, "lr": 3.487866168517775e-05, "epoch": 1.3013100436681222, "percentage": 43.38, "elapsed_time": "11:50:41", "remaining_time": "15:27:42"} +{"current_steps": 2985, "total_steps": 6870, "loss": 0.2819, "lr": 3.4820285942330003e-05, "epoch": 1.3034934497816595, "percentage": 43.45, "elapsed_time": "11:51:52", "remaining_time": "15:26:30"} +{"current_steps": 2990, "total_steps": 6870, "loss": 0.2801, "lr": 3.476184681759581e-05, "epoch": 1.3056768558951966, "percentage": 43.52, "elapsed_time": "11:53:04", "remaining_time": "15:25:19"} +{"current_steps": 2995, "total_steps": 6870, "loss": 0.2783, "lr": 3.4703344688151776e-05, "epoch": 1.3078602620087336, "percentage": 43.6, "elapsed_time": "11:54:15", "remaining_time": "15:24:07"} +{"current_steps": 3000, "total_steps": 6870, "loss": 0.2818, "lr": 3.46447799315811e-05, "epoch": 1.3100436681222707, "percentage": 43.67, "elapsed_time": "11:55:27", "remaining_time": "15:22:56"} +{"current_steps": 3005, "total_steps": 6870, "loss": 0.279, "lr": 3.458615292587125e-05, "epoch": 1.3122270742358078, "percentage": 43.74, "elapsed_time": "11:56:38", "remaining_time": "15:21:44"} +{"current_steps": 3010, "total_steps": 6870, "loss": 0.2858, "lr": 3.452746404941143e-05, "epoch": 1.314410480349345, "percentage": 43.81, "elapsed_time": "11:57:50", "remaining_time": "15:20:33"} +{"current_steps": 3015, "total_steps": 6870, "loss": 0.2747, "lr": 3.446871368099017e-05, "epoch": 1.3165938864628821, "percentage": 43.89, "elapsed_time": "11:59:02", "remaining_time": "15:19:21"} +{"current_steps": 3020, "total_steps": 6870, "loss": 0.2862, "lr": 3.440990219979289e-05, "epoch": 1.3187772925764192, "percentage": 43.96, "elapsed_time": "12:00:13", "remaining_time": "15:18:10"} +{"current_steps": 3025, "total_steps": 6870, "loss": 0.2703, "lr": 3.435102998539942e-05, "epoch": 1.3209606986899565, "percentage": 44.03, "elapsed_time": "12:01:25", "remaining_time": "15:16:58"} +{"current_steps": 3030, "total_steps": 6870, "loss": 0.2695, "lr": 3.4292097417781595e-05, "epoch": 1.3231441048034935, "percentage": 44.1, "elapsed_time": "12:02:36", "remaining_time": "15:15:47"} +{"current_steps": 3035, "total_steps": 6870, "loss": 0.2666, "lr": 3.4233104877300775e-05, "epoch": 1.3253275109170306, "percentage": 44.18, "elapsed_time": "12:03:48", "remaining_time": "15:14:35"} +{"current_steps": 3040, "total_steps": 6870, "loss": 0.2658, "lr": 3.41740527447054e-05, "epoch": 1.3275109170305677, "percentage": 44.25, "elapsed_time": "12:04:59", "remaining_time": "15:13:24"} +{"current_steps": 3045, "total_steps": 6870, "loss": 0.2716, "lr": 3.411494140112852e-05, "epoch": 1.3296943231441047, "percentage": 44.32, "elapsed_time": "12:06:11", "remaining_time": "15:12:12"} +{"current_steps": 3050, "total_steps": 6870, "loss": 0.288, "lr": 3.4055771228085347e-05, "epoch": 1.3318777292576418, "percentage": 44.4, "elapsed_time": "12:07:22", "remaining_time": "15:11:00"} +{"current_steps": 3055, "total_steps": 6870, "loss": 0.2723, "lr": 3.399654260747078e-05, "epoch": 1.334061135371179, "percentage": 44.47, "elapsed_time": "12:08:34", "remaining_time": "15:09:49"} +{"current_steps": 3060, "total_steps": 6870, "loss": 0.2877, "lr": 3.3937255921556974e-05, "epoch": 1.3362445414847162, "percentage": 44.54, "elapsed_time": "12:09:46", "remaining_time": "15:08:37"} +{"current_steps": 3065, "total_steps": 6870, "loss": 0.2782, "lr": 3.3877911552990826e-05, "epoch": 1.3384279475982532, "percentage": 44.61, "elapsed_time": "12:10:57", "remaining_time": "15:07:26"} +{"current_steps": 3070, "total_steps": 6870, "loss": 0.2727, "lr": 3.3818509884791514e-05, "epoch": 1.3406113537117905, "percentage": 44.69, "elapsed_time": "12:12:09", "remaining_time": "15:06:14"} +{"current_steps": 3075, "total_steps": 6870, "loss": 0.2741, "lr": 3.375905130034808e-05, "epoch": 1.3427947598253276, "percentage": 44.76, "elapsed_time": "12:13:20", "remaining_time": "15:05:03"} +{"current_steps": 3080, "total_steps": 6870, "loss": 0.287, "lr": 3.369953618341688e-05, "epoch": 1.3449781659388647, "percentage": 44.83, "elapsed_time": "12:14:32", "remaining_time": "15:03:51"} +{"current_steps": 3085, "total_steps": 6870, "loss": 0.2753, "lr": 3.3639964918119135e-05, "epoch": 1.3471615720524017, "percentage": 44.91, "elapsed_time": "12:15:43", "remaining_time": "15:02:40"} +{"current_steps": 3090, "total_steps": 6870, "loss": 0.2795, "lr": 3.3580337888938515e-05, "epoch": 1.3493449781659388, "percentage": 44.98, "elapsed_time": "12:16:55", "remaining_time": "15:01:28"} +{"current_steps": 3095, "total_steps": 6870, "loss": 0.2818, "lr": 3.3520655480718515e-05, "epoch": 1.3515283842794759, "percentage": 45.05, "elapsed_time": "12:18:07", "remaining_time": "15:00:17"} +{"current_steps": 3100, "total_steps": 6870, "loss": 0.2799, "lr": 3.346091807866013e-05, "epoch": 1.3537117903930131, "percentage": 45.12, "elapsed_time": "12:19:18", "remaining_time": "14:59:05"} +{"current_steps": 3105, "total_steps": 6870, "loss": 0.2748, "lr": 3.3401126068319244e-05, "epoch": 1.3558951965065502, "percentage": 45.2, "elapsed_time": "12:20:30", "remaining_time": "14:57:54"} +{"current_steps": 3110, "total_steps": 6870, "loss": 0.2799, "lr": 3.3341279835604224e-05, "epoch": 1.3580786026200873, "percentage": 45.27, "elapsed_time": "12:21:41", "remaining_time": "14:56:42"} +{"current_steps": 3115, "total_steps": 6870, "loss": 0.2948, "lr": 3.328137976677339e-05, "epoch": 1.3602620087336246, "percentage": 45.34, "elapsed_time": "12:22:53", "remaining_time": "14:55:30"} +{"current_steps": 3120, "total_steps": 6870, "loss": 0.2883, "lr": 3.322142624843253e-05, "epoch": 1.3624454148471616, "percentage": 45.41, "elapsed_time": "12:24:04", "remaining_time": "14:54:19"} +{"current_steps": 3125, "total_steps": 6870, "loss": 0.2813, "lr": 3.3161419667532384e-05, "epoch": 1.3646288209606987, "percentage": 45.49, "elapsed_time": "12:25:16", "remaining_time": "14:53:07"} +{"current_steps": 3130, "total_steps": 6870, "loss": 0.2714, "lr": 3.3101360411366205e-05, "epoch": 1.3668122270742358, "percentage": 45.56, "elapsed_time": "12:26:27", "remaining_time": "14:51:56"} +{"current_steps": 3135, "total_steps": 6870, "loss": 0.2773, "lr": 3.304124886756719e-05, "epoch": 1.3689956331877728, "percentage": 45.63, "elapsed_time": "12:27:39", "remaining_time": "14:50:44"} +{"current_steps": 3140, "total_steps": 6870, "loss": 0.2718, "lr": 3.2981085424106036e-05, "epoch": 1.37117903930131, "percentage": 45.71, "elapsed_time": "12:28:50", "remaining_time": "14:49:33"} +{"current_steps": 3145, "total_steps": 6870, "loss": 0.293, "lr": 3.2920870469288375e-05, "epoch": 1.3733624454148472, "percentage": 45.78, "elapsed_time": "12:30:02", "remaining_time": "14:48:21"} +{"current_steps": 3150, "total_steps": 6870, "loss": 0.2767, "lr": 3.2860604391752344e-05, "epoch": 1.3755458515283843, "percentage": 45.85, "elapsed_time": "12:31:13", "remaining_time": "14:47:09"} +{"current_steps": 3155, "total_steps": 6870, "loss": 0.28, "lr": 3.2800287580466e-05, "epoch": 1.3777292576419213, "percentage": 45.92, "elapsed_time": "12:32:25", "remaining_time": "14:45:58"} +{"current_steps": 3160, "total_steps": 6870, "loss": 0.2626, "lr": 3.273992042472487e-05, "epoch": 1.3799126637554586, "percentage": 46.0, "elapsed_time": "12:33:36", "remaining_time": "14:44:46"} +{"current_steps": 3165, "total_steps": 6870, "loss": 0.2823, "lr": 3.2679503314149395e-05, "epoch": 1.3820960698689957, "percentage": 46.07, "elapsed_time": "12:34:48", "remaining_time": "14:43:35"} +{"current_steps": 3170, "total_steps": 6870, "loss": 0.2774, "lr": 3.261903663868243e-05, "epoch": 1.3842794759825328, "percentage": 46.14, "elapsed_time": "12:35:59", "remaining_time": "14:42:23"} +{"current_steps": 3175, "total_steps": 6870, "loss": 0.2706, "lr": 3.2558520788586764e-05, "epoch": 1.3864628820960698, "percentage": 46.22, "elapsed_time": "12:37:11", "remaining_time": "14:41:12"} +{"current_steps": 3180, "total_steps": 6870, "loss": 0.2652, "lr": 3.249795615444253e-05, "epoch": 1.388646288209607, "percentage": 46.29, "elapsed_time": "12:38:23", "remaining_time": "14:40:00"} +{"current_steps": 3185, "total_steps": 6870, "loss": 0.277, "lr": 3.243734312714475e-05, "epoch": 1.3908296943231442, "percentage": 46.36, "elapsed_time": "12:39:34", "remaining_time": "14:38:49"} +{"current_steps": 3190, "total_steps": 6870, "loss": 0.2796, "lr": 3.2376682097900754e-05, "epoch": 1.3930131004366813, "percentage": 46.43, "elapsed_time": "12:40:46", "remaining_time": "14:37:37"} +{"current_steps": 3195, "total_steps": 6870, "loss": 0.2875, "lr": 3.2315973458227714e-05, "epoch": 1.3951965065502183, "percentage": 46.51, "elapsed_time": "12:41:57", "remaining_time": "14:36:26"} +{"current_steps": 3200, "total_steps": 6870, "loss": 0.2817, "lr": 3.225521759995007e-05, "epoch": 1.3973799126637554, "percentage": 46.58, "elapsed_time": "12:43:09", "remaining_time": "14:35:14"} +{"current_steps": 3205, "total_steps": 6870, "loss": 0.2742, "lr": 3.219441491519703e-05, "epoch": 1.3995633187772927, "percentage": 46.65, "elapsed_time": "12:44:20", "remaining_time": "14:34:03"} +{"current_steps": 3210, "total_steps": 6870, "loss": 0.2791, "lr": 3.2133565796400046e-05, "epoch": 1.4017467248908297, "percentage": 46.72, "elapsed_time": "12:45:32", "remaining_time": "14:32:51"} +{"current_steps": 3215, "total_steps": 6870, "loss": 0.2687, "lr": 3.2072670636290195e-05, "epoch": 1.4039301310043668, "percentage": 46.8, "elapsed_time": "12:46:43", "remaining_time": "14:31:40"} +{"current_steps": 3220, "total_steps": 6870, "loss": 0.2743, "lr": 3.20117298278958e-05, "epoch": 1.4061135371179039, "percentage": 46.87, "elapsed_time": "12:47:55", "remaining_time": "14:30:28"} +{"current_steps": 3225, "total_steps": 6870, "loss": 0.2904, "lr": 3.195074376453975e-05, "epoch": 1.408296943231441, "percentage": 46.94, "elapsed_time": "12:49:07", "remaining_time": "14:29:16"} +{"current_steps": 3230, "total_steps": 6870, "loss": 0.2635, "lr": 3.188971283983703e-05, "epoch": 1.4104803493449782, "percentage": 47.02, "elapsed_time": "12:50:18", "remaining_time": "14:28:05"} +{"current_steps": 3235, "total_steps": 6870, "loss": 0.2752, "lr": 3.1828637447692187e-05, "epoch": 1.4126637554585153, "percentage": 47.09, "elapsed_time": "12:51:30", "remaining_time": "14:26:53"} +{"current_steps": 3240, "total_steps": 6870, "loss": 0.2802, "lr": 3.176751798229673e-05, "epoch": 1.4148471615720524, "percentage": 47.16, "elapsed_time": "12:52:41", "remaining_time": "14:25:42"} +{"current_steps": 3245, "total_steps": 6870, "loss": 0.2583, "lr": 3.1706354838126687e-05, "epoch": 1.4170305676855894, "percentage": 47.23, "elapsed_time": "12:53:53", "remaining_time": "14:24:30"} +{"current_steps": 3250, "total_steps": 6870, "loss": 0.2747, "lr": 3.1645148409939915e-05, "epoch": 1.4192139737991267, "percentage": 47.31, "elapsed_time": "12:55:04", "remaining_time": "14:23:19"} +{"current_steps": 3255, "total_steps": 6870, "loss": 0.2857, "lr": 3.158389909277371e-05, "epoch": 1.4213973799126638, "percentage": 47.38, "elapsed_time": "12:56:16", "remaining_time": "14:22:07"} +{"current_steps": 3260, "total_steps": 6870, "loss": 0.2824, "lr": 3.152260728194213e-05, "epoch": 1.4235807860262009, "percentage": 47.45, "elapsed_time": "12:57:27", "remaining_time": "14:20:56"} +{"current_steps": 3265, "total_steps": 6870, "loss": 0.2789, "lr": 3.1461273373033516e-05, "epoch": 1.425764192139738, "percentage": 47.53, "elapsed_time": "12:58:39", "remaining_time": "14:19:44"} +{"current_steps": 3270, "total_steps": 6870, "loss": 0.2783, "lr": 3.1399897761907914e-05, "epoch": 1.427947598253275, "percentage": 47.6, "elapsed_time": "12:59:50", "remaining_time": "14:18:32"} +{"current_steps": 3275, "total_steps": 6870, "loss": 0.272, "lr": 3.133848084469454e-05, "epoch": 1.4301310043668123, "percentage": 47.67, "elapsed_time": "13:01:02", "remaining_time": "14:17:21"} +{"current_steps": 3280, "total_steps": 6870, "loss": 0.2694, "lr": 3.1277023017789164e-05, "epoch": 1.4323144104803494, "percentage": 47.74, "elapsed_time": "13:02:14", "remaining_time": "14:16:09"} +{"current_steps": 3285, "total_steps": 6870, "loss": 0.2756, "lr": 3.121552467785164e-05, "epoch": 1.4344978165938864, "percentage": 47.82, "elapsed_time": "13:03:25", "remaining_time": "14:14:58"} +{"current_steps": 3290, "total_steps": 6870, "loss": 0.2631, "lr": 3.1153986221803284e-05, "epoch": 1.4366812227074237, "percentage": 47.89, "elapsed_time": "13:04:37", "remaining_time": "14:13:46"} +{"current_steps": 3295, "total_steps": 6870, "loss": 0.2747, "lr": 3.109240804682431e-05, "epoch": 1.4388646288209608, "percentage": 47.96, "elapsed_time": "13:05:48", "remaining_time": "14:12:35"} +{"current_steps": 3300, "total_steps": 6870, "loss": 0.2735, "lr": 3.103079055035131e-05, "epoch": 1.4410480349344978, "percentage": 48.03, "elapsed_time": "13:07:00", "remaining_time": "14:11:23"} +{"current_steps": 3305, "total_steps": 6870, "loss": 0.2718, "lr": 3.096913413007467e-05, "epoch": 1.443231441048035, "percentage": 48.11, "elapsed_time": "13:08:11", "remaining_time": "14:10:12"} +{"current_steps": 3310, "total_steps": 6870, "loss": 0.2713, "lr": 3.090743918393596e-05, "epoch": 1.445414847161572, "percentage": 48.18, "elapsed_time": "13:09:23", "remaining_time": "14:09:00"} +{"current_steps": 3315, "total_steps": 6870, "loss": 0.2753, "lr": 3.0845706110125435e-05, "epoch": 1.447598253275109, "percentage": 48.25, "elapsed_time": "13:10:35", "remaining_time": "14:07:49"} +{"current_steps": 3320, "total_steps": 6870, "loss": 0.2706, "lr": 3.078393530707942e-05, "epoch": 1.4497816593886463, "percentage": 48.33, "elapsed_time": "13:11:46", "remaining_time": "14:06:37"} +{"current_steps": 3325, "total_steps": 6870, "loss": 0.2797, "lr": 3.072212717347776e-05, "epoch": 1.4519650655021834, "percentage": 48.4, "elapsed_time": "13:12:58", "remaining_time": "14:05:26"} +{"current_steps": 3330, "total_steps": 6870, "loss": 0.2793, "lr": 3.0660282108241235e-05, "epoch": 1.4541484716157205, "percentage": 48.47, "elapsed_time": "13:14:09", "remaining_time": "14:04:14"} +{"current_steps": 3335, "total_steps": 6870, "loss": 0.2911, "lr": 3.059840051052898e-05, "epoch": 1.4563318777292578, "percentage": 48.54, "elapsed_time": "13:15:21", "remaining_time": "14:03:03"} +{"current_steps": 3340, "total_steps": 6870, "loss": 0.267, "lr": 3.053648277973592e-05, "epoch": 1.4585152838427948, "percentage": 48.62, "elapsed_time": "13:16:33", "remaining_time": "14:01:51"} +{"current_steps": 3345, "total_steps": 6870, "loss": 0.2858, "lr": 3.04745293154902e-05, "epoch": 1.460698689956332, "percentage": 48.69, "elapsed_time": "13:17:44", "remaining_time": "14:00:40"} +{"current_steps": 3350, "total_steps": 6870, "loss": 0.2683, "lr": 3.0412540517650582e-05, "epoch": 1.462882096069869, "percentage": 48.76, "elapsed_time": "13:18:56", "remaining_time": "13:59:28"} +{"current_steps": 3355, "total_steps": 6870, "loss": 0.2859, "lr": 3.03505167863039e-05, "epoch": 1.465065502183406, "percentage": 48.84, "elapsed_time": "13:20:07", "remaining_time": "13:58:17"} +{"current_steps": 3360, "total_steps": 6870, "loss": 0.2804, "lr": 3.0288458521762414e-05, "epoch": 1.467248908296943, "percentage": 48.91, "elapsed_time": "13:21:19", "remaining_time": "13:57:05"} +{"current_steps": 3365, "total_steps": 6870, "loss": 0.2671, "lr": 3.02263661245613e-05, "epoch": 1.4694323144104804, "percentage": 48.98, "elapsed_time": "13:22:30", "remaining_time": "13:55:54"} +{"current_steps": 3370, "total_steps": 6870, "loss": 0.2807, "lr": 3.016423999545603e-05, "epoch": 1.4716157205240175, "percentage": 49.05, "elapsed_time": "13:23:42", "remaining_time": "13:54:42"} +{"current_steps": 3375, "total_steps": 6870, "loss": 0.2846, "lr": 3.0102080535419775e-05, "epoch": 1.4737991266375545, "percentage": 49.13, "elapsed_time": "13:24:54", "remaining_time": "13:53:31"} +{"current_steps": 3380, "total_steps": 6870, "loss": 0.2768, "lr": 3.0039888145640845e-05, "epoch": 1.4759825327510918, "percentage": 49.2, "elapsed_time": "13:26:05", "remaining_time": "13:52:19"} +{"current_steps": 3385, "total_steps": 6870, "loss": 0.2732, "lr": 2.9977663227520065e-05, "epoch": 1.4781659388646289, "percentage": 49.27, "elapsed_time": "13:27:17", "remaining_time": "13:51:08"} +{"current_steps": 3390, "total_steps": 6870, "loss": 0.2692, "lr": 2.991540618266822e-05, "epoch": 1.480349344978166, "percentage": 49.34, "elapsed_time": "13:28:28", "remaining_time": "13:49:56"} +{"current_steps": 3395, "total_steps": 6870, "loss": 0.259, "lr": 2.9853117412903443e-05, "epoch": 1.482532751091703, "percentage": 49.42, "elapsed_time": "13:29:40", "remaining_time": "13:48:45"} +{"current_steps": 3400, "total_steps": 6870, "loss": 0.2674, "lr": 2.979079732024862e-05, "epoch": 1.48471615720524, "percentage": 49.49, "elapsed_time": "13:30:51", "remaining_time": "13:47:33"} +{"current_steps": 3405, "total_steps": 6870, "loss": 0.2741, "lr": 2.9728446306928813e-05, "epoch": 1.4868995633187774, "percentage": 49.56, "elapsed_time": "13:32:03", "remaining_time": "13:46:21"} +{"current_steps": 3410, "total_steps": 6870, "loss": 0.2648, "lr": 2.966606477536863e-05, "epoch": 1.4890829694323144, "percentage": 49.64, "elapsed_time": "13:33:14", "remaining_time": "13:45:10"} +{"current_steps": 3415, "total_steps": 6870, "loss": 0.2743, "lr": 2.960365312818967e-05, "epoch": 1.4912663755458515, "percentage": 49.71, "elapsed_time": "13:34:26", "remaining_time": "13:43:58"} +{"current_steps": 3420, "total_steps": 6870, "loss": 0.2716, "lr": 2.9541211768207883e-05, "epoch": 1.4934497816593886, "percentage": 49.78, "elapsed_time": "13:35:37", "remaining_time": "13:42:47"} +{"current_steps": 3425, "total_steps": 6870, "loss": 0.2748, "lr": 2.9478741098431027e-05, "epoch": 1.4956331877729259, "percentage": 49.85, "elapsed_time": "13:36:49", "remaining_time": "13:41:35"} +{"current_steps": 3430, "total_steps": 6870, "loss": 0.301, "lr": 2.9416241522055987e-05, "epoch": 1.497816593886463, "percentage": 49.93, "elapsed_time": "13:38:01", "remaining_time": "13:40:24"} +{"current_steps": 3435, "total_steps": 6870, "loss": 0.2799, "lr": 2.9353713442466246e-05, "epoch": 1.5, "percentage": 50.0, "elapsed_time": "13:39:12", "remaining_time": "13:39:12"} +{"current_steps": 3440, "total_steps": 6870, "loss": 0.2712, "lr": 2.929115726322924e-05, "epoch": 1.502183406113537, "percentage": 50.07, "elapsed_time": "13:40:24", "remaining_time": "13:38:01"} +{"current_steps": 3445, "total_steps": 6870, "loss": 0.2778, "lr": 2.9228573388093765e-05, "epoch": 1.5043668122270741, "percentage": 50.15, "elapsed_time": "13:41:35", "remaining_time": "13:36:49"} +{"current_steps": 3450, "total_steps": 6870, "loss": 0.2779, "lr": 2.916596222098738e-05, "epoch": 1.5065502183406112, "percentage": 50.22, "elapsed_time": "13:42:47", "remaining_time": "13:35:37"} +{"current_steps": 3455, "total_steps": 6870, "loss": 0.272, "lr": 2.9103324166013777e-05, "epoch": 1.5087336244541485, "percentage": 50.29, "elapsed_time": "13:43:58", "remaining_time": "13:34:26"} +{"current_steps": 3460, "total_steps": 6870, "loss": 0.2732, "lr": 2.9040659627450195e-05, "epoch": 1.5109170305676856, "percentage": 50.36, "elapsed_time": "13:45:10", "remaining_time": "13:33:14"} +{"current_steps": 3465, "total_steps": 6870, "loss": 0.2694, "lr": 2.897796900974481e-05, "epoch": 1.5131004366812228, "percentage": 50.44, "elapsed_time": "13:46:21", "remaining_time": "13:32:03"} +{"current_steps": 3470, "total_steps": 6870, "loss": 0.2752, "lr": 2.8915252717514108e-05, "epoch": 1.51528384279476, "percentage": 50.51, "elapsed_time": "13:47:33", "remaining_time": "13:30:51"} +{"current_steps": 3475, "total_steps": 6870, "loss": 0.28, "lr": 2.8852511155540286e-05, "epoch": 1.517467248908297, "percentage": 50.58, "elapsed_time": "13:48:45", "remaining_time": "13:29:40"} +{"current_steps": 3480, "total_steps": 6870, "loss": 0.2669, "lr": 2.8789744728768632e-05, "epoch": 1.519650655021834, "percentage": 50.66, "elapsed_time": "13:49:56", "remaining_time": "13:28:28"} +{"current_steps": 3485, "total_steps": 6870, "loss": 0.2774, "lr": 2.8726953842304917e-05, "epoch": 1.5218340611353711, "percentage": 50.73, "elapsed_time": "13:51:08", "remaining_time": "13:27:17"} +{"current_steps": 3490, "total_steps": 6870, "loss": 0.2822, "lr": 2.8664138901412784e-05, "epoch": 1.5240174672489082, "percentage": 50.8, "elapsed_time": "13:52:19", "remaining_time": "13:26:05"} +{"current_steps": 3495, "total_steps": 6870, "loss": 0.2701, "lr": 2.8601300311511135e-05, "epoch": 1.5262008733624453, "percentage": 50.87, "elapsed_time": "13:53:31", "remaining_time": "13:24:54"} +{"current_steps": 3500, "total_steps": 6870, "loss": 0.2705, "lr": 2.8538438478171493e-05, "epoch": 1.5283842794759825, "percentage": 50.95, "elapsed_time": "13:54:42", "remaining_time": "13:23:42"} +{"current_steps": 3505, "total_steps": 6870, "loss": 0.2645, "lr": 2.8475553807115392e-05, "epoch": 1.5305676855895196, "percentage": 51.02, "elapsed_time": "13:55:54", "remaining_time": "13:22:31"} +{"current_steps": 3510, "total_steps": 6870, "loss": 0.2716, "lr": 2.8412646704211787e-05, "epoch": 1.532751091703057, "percentage": 51.09, "elapsed_time": "13:57:05", "remaining_time": "13:21:19"} +{"current_steps": 3515, "total_steps": 6870, "loss": 0.281, "lr": 2.8349717575474395e-05, "epoch": 1.534934497816594, "percentage": 51.16, "elapsed_time": "13:58:17", "remaining_time": "13:20:07"} +{"current_steps": 3520, "total_steps": 6870, "loss": 0.285, "lr": 2.8286766827059103e-05, "epoch": 1.537117903930131, "percentage": 51.24, "elapsed_time": "13:59:28", "remaining_time": "13:18:56"} +{"current_steps": 3525, "total_steps": 6870, "loss": 0.2705, "lr": 2.8223794865261326e-05, "epoch": 1.539301310043668, "percentage": 51.31, "elapsed_time": "14:00:40", "remaining_time": "13:17:44"} +{"current_steps": 3530, "total_steps": 6870, "loss": 0.2643, "lr": 2.8160802096513396e-05, "epoch": 1.5414847161572052, "percentage": 51.38, "elapsed_time": "14:01:51", "remaining_time": "13:16:33"} +{"current_steps": 3535, "total_steps": 6870, "loss": 0.2797, "lr": 2.809778892738194e-05, "epoch": 1.5436681222707422, "percentage": 51.46, "elapsed_time": "14:03:03", "remaining_time": "13:15:21"} +{"current_steps": 3540, "total_steps": 6870, "loss": 0.2702, "lr": 2.8034755764565252e-05, "epoch": 1.5458515283842795, "percentage": 51.53, "elapsed_time": "14:04:15", "remaining_time": "13:14:10"} +{"current_steps": 3545, "total_steps": 6870, "loss": 0.2717, "lr": 2.797170301489066e-05, "epoch": 1.5480349344978166, "percentage": 51.6, "elapsed_time": "14:05:26", "remaining_time": "13:12:58"} +{"current_steps": 3550, "total_steps": 6870, "loss": 0.2618, "lr": 2.7908631085311933e-05, "epoch": 1.5502183406113537, "percentage": 51.67, "elapsed_time": "14:06:38", "remaining_time": "13:11:47"} +{"current_steps": 3555, "total_steps": 6870, "loss": 0.2638, "lr": 2.78455403829066e-05, "epoch": 1.552401746724891, "percentage": 51.75, "elapsed_time": "14:07:49", "remaining_time": "13:10:35"} +{"current_steps": 3560, "total_steps": 6870, "loss": 0.2681, "lr": 2.7782431314873377e-05, "epoch": 1.554585152838428, "percentage": 51.82, "elapsed_time": "14:09:01", "remaining_time": "13:09:24"} +{"current_steps": 3565, "total_steps": 6870, "loss": 0.2715, "lr": 2.7719304288529503e-05, "epoch": 1.556768558951965, "percentage": 51.89, "elapsed_time": "14:10:13", "remaining_time": "13:08:12"} +{"current_steps": 3570, "total_steps": 6870, "loss": 0.2755, "lr": 2.7656159711308127e-05, "epoch": 1.5589519650655022, "percentage": 51.97, "elapsed_time": "14:11:24", "remaining_time": "13:07:01"} +{"current_steps": 3575, "total_steps": 6870, "loss": 0.267, "lr": 2.7592997990755665e-05, "epoch": 1.5611353711790392, "percentage": 52.04, "elapsed_time": "14:12:36", "remaining_time": "13:05:49"} +{"current_steps": 3580, "total_steps": 6870, "loss": 0.2691, "lr": 2.75298195345292e-05, "epoch": 1.5633187772925763, "percentage": 52.11, "elapsed_time": "14:13:47", "remaining_time": "13:04:38"} +{"current_steps": 3585, "total_steps": 6870, "loss": 0.2619, "lr": 2.7466624750393804e-05, "epoch": 1.5655021834061136, "percentage": 52.18, "elapsed_time": "14:14:59", "remaining_time": "13:03:26"} +{"current_steps": 3590, "total_steps": 6870, "loss": 0.2703, "lr": 2.740341404621996e-05, "epoch": 1.5676855895196506, "percentage": 52.26, "elapsed_time": "14:16:10", "remaining_time": "13:02:15"} +{"current_steps": 3595, "total_steps": 6870, "loss": 0.2594, "lr": 2.7340187829980884e-05, "epoch": 1.569868995633188, "percentage": 52.33, "elapsed_time": "14:17:22", "remaining_time": "13:01:03"} +{"current_steps": 3600, "total_steps": 6870, "loss": 0.2754, "lr": 2.7276946509749905e-05, "epoch": 1.572052401746725, "percentage": 52.4, "elapsed_time": "14:18:34", "remaining_time": "12:59:51"} +{"current_steps": 3605, "total_steps": 6870, "loss": 0.2887, "lr": 2.7213690493697852e-05, "epoch": 1.574235807860262, "percentage": 52.47, "elapsed_time": "14:19:45", "remaining_time": "12:58:40"} +{"current_steps": 3610, "total_steps": 6870, "loss": 0.2661, "lr": 2.7150420190090397e-05, "epoch": 1.5764192139737991, "percentage": 52.55, "elapsed_time": "14:20:57", "remaining_time": "12:57:28"} +{"current_steps": 3615, "total_steps": 6870, "loss": 0.2698, "lr": 2.7087136007285417e-05, "epoch": 1.5786026200873362, "percentage": 52.62, "elapsed_time": "14:22:08", "remaining_time": "12:56:17"} +{"current_steps": 3620, "total_steps": 6870, "loss": 0.2603, "lr": 2.7023838353730386e-05, "epoch": 1.5807860262008733, "percentage": 52.69, "elapsed_time": "14:23:20", "remaining_time": "12:55:05"} +{"current_steps": 3625, "total_steps": 6870, "loss": 0.2609, "lr": 2.6960527637959704e-05, "epoch": 1.5829694323144103, "percentage": 52.77, "elapsed_time": "14:24:31", "remaining_time": "12:53:54"} +{"current_steps": 3630, "total_steps": 6870, "loss": 0.2743, "lr": 2.689720426859208e-05, "epoch": 1.5851528384279476, "percentage": 52.84, "elapsed_time": "14:25:43", "remaining_time": "12:52:42"} +{"current_steps": 3635, "total_steps": 6870, "loss": 0.2754, "lr": 2.68338686543279e-05, "epoch": 1.5873362445414847, "percentage": 52.91, "elapsed_time": "14:26:54", "remaining_time": "12:51:31"} +{"current_steps": 3640, "total_steps": 6870, "loss": 0.2689, "lr": 2.6770521203946582e-05, "epoch": 1.589519650655022, "percentage": 52.98, "elapsed_time": "14:28:06", "remaining_time": "12:50:19"} +{"current_steps": 3645, "total_steps": 6870, "loss": 0.2709, "lr": 2.670716232630392e-05, "epoch": 1.591703056768559, "percentage": 53.06, "elapsed_time": "14:29:18", "remaining_time": "12:49:08"} +{"current_steps": 3650, "total_steps": 6870, "loss": 0.269, "lr": 2.6643792430329473e-05, "epoch": 1.5938864628820961, "percentage": 53.13, "elapsed_time": "14:30:29", "remaining_time": "12:47:56"} +{"current_steps": 3655, "total_steps": 6870, "loss": 0.2891, "lr": 2.6580411925023916e-05, "epoch": 1.5960698689956332, "percentage": 53.2, "elapsed_time": "14:31:41", "remaining_time": "12:46:44"} +{"current_steps": 3660, "total_steps": 6870, "loss": 0.2771, "lr": 2.6517021219456402e-05, "epoch": 1.5982532751091703, "percentage": 53.28, "elapsed_time": "14:32:52", "remaining_time": "12:45:33"} +{"current_steps": 3665, "total_steps": 6870, "loss": 0.2767, "lr": 2.6453620722761896e-05, "epoch": 1.6004366812227073, "percentage": 53.35, "elapsed_time": "14:34:04", "remaining_time": "12:44:21"} +{"current_steps": 3670, "total_steps": 6870, "loss": 0.2772, "lr": 2.6390210844138593e-05, "epoch": 1.6026200873362444, "percentage": 53.42, "elapsed_time": "14:35:15", "remaining_time": "12:43:10"} +{"current_steps": 3675, "total_steps": 6870, "loss": 0.2748, "lr": 2.6326791992845202e-05, "epoch": 1.6048034934497817, "percentage": 53.49, "elapsed_time": "14:36:27", "remaining_time": "12:41:58"} +{"current_steps": 3680, "total_steps": 6870, "loss": 0.2629, "lr": 2.626336457819837e-05, "epoch": 1.6069868995633187, "percentage": 53.57, "elapsed_time": "14:37:38", "remaining_time": "12:40:47"} +{"current_steps": 3685, "total_steps": 6870, "loss": 0.2897, "lr": 2.619992900957e-05, "epoch": 1.609170305676856, "percentage": 53.64, "elapsed_time": "14:38:50", "remaining_time": "12:39:35"} +{"current_steps": 3690, "total_steps": 6870, "loss": 0.283, "lr": 2.613648569638464e-05, "epoch": 1.611353711790393, "percentage": 53.71, "elapsed_time": "14:40:01", "remaining_time": "12:38:24"} +{"current_steps": 3695, "total_steps": 6870, "loss": 0.2691, "lr": 2.6073035048116796e-05, "epoch": 1.6135371179039302, "percentage": 53.78, "elapsed_time": "14:41:13", "remaining_time": "12:37:12"} +{"current_steps": 3700, "total_steps": 6870, "loss": 0.2759, "lr": 2.6009577474288333e-05, "epoch": 1.6157205240174672, "percentage": 53.86, "elapsed_time": "14:42:24", "remaining_time": "12:36:00"} +{"current_steps": 3705, "total_steps": 6870, "loss": 0.2639, "lr": 2.5946113384465813e-05, "epoch": 1.6179039301310043, "percentage": 53.93, "elapsed_time": "14:43:36", "remaining_time": "12:34:49"} +{"current_steps": 3710, "total_steps": 6870, "loss": 0.2706, "lr": 2.5882643188257856e-05, "epoch": 1.6200873362445414, "percentage": 54.0, "elapsed_time": "14:44:48", "remaining_time": "12:33:37"} +{"current_steps": 3715, "total_steps": 6870, "loss": 0.2722, "lr": 2.5819167295312485e-05, "epoch": 1.6222707423580784, "percentage": 54.08, "elapsed_time": "14:45:59", "remaining_time": "12:32:26"} +{"current_steps": 3720, "total_steps": 6870, "loss": 0.2705, "lr": 2.57556861153145e-05, "epoch": 1.6244541484716157, "percentage": 54.15, "elapsed_time": "14:47:11", "remaining_time": "12:31:14"} +{"current_steps": 3725, "total_steps": 6870, "loss": 0.2724, "lr": 2.569220005798281e-05, "epoch": 1.6266375545851528, "percentage": 54.22, "elapsed_time": "14:48:22", "remaining_time": "12:30:03"} +{"current_steps": 3730, "total_steps": 6870, "loss": 0.2713, "lr": 2.5628709533067824e-05, "epoch": 1.62882096069869, "percentage": 54.29, "elapsed_time": "14:49:34", "remaining_time": "12:28:51"} +{"current_steps": 3735, "total_steps": 6870, "loss": 0.2605, "lr": 2.556521495034877e-05, "epoch": 1.6310043668122272, "percentage": 54.37, "elapsed_time": "14:50:46", "remaining_time": "12:27:40"} +{"current_steps": 3740, "total_steps": 6870, "loss": 0.2571, "lr": 2.550171671963108e-05, "epoch": 1.6331877729257642, "percentage": 54.44, "elapsed_time": "14:51:57", "remaining_time": "12:26:28"} +{"current_steps": 3745, "total_steps": 6870, "loss": 0.2696, "lr": 2.543821525074371e-05, "epoch": 1.6353711790393013, "percentage": 54.51, "elapsed_time": "14:53:09", "remaining_time": "12:25:17"} +{"current_steps": 3750, "total_steps": 6870, "loss": 0.2601, "lr": 2.537471095353653e-05, "epoch": 1.6375545851528384, "percentage": 54.59, "elapsed_time": "14:54:20", "remaining_time": "12:24:05"} +{"current_steps": 3755, "total_steps": 6870, "loss": 0.2711, "lr": 2.5311204237877657e-05, "epoch": 1.6397379912663754, "percentage": 54.66, "elapsed_time": "14:55:32", "remaining_time": "12:22:54"} +{"current_steps": 3760, "total_steps": 6870, "loss": 0.2565, "lr": 2.524769551365083e-05, "epoch": 1.6419213973799127, "percentage": 54.73, "elapsed_time": "14:56:43", "remaining_time": "12:21:42"} +{"current_steps": 3765, "total_steps": 6870, "loss": 0.2804, "lr": 2.5184185190752752e-05, "epoch": 1.6441048034934498, "percentage": 54.8, "elapsed_time": "14:57:55", "remaining_time": "12:20:30"} +{"current_steps": 3770, "total_steps": 6870, "loss": 0.2722, "lr": 2.512067367909041e-05, "epoch": 1.6462882096069869, "percentage": 54.88, "elapsed_time": "14:59:06", "remaining_time": "12:19:19"} +{"current_steps": 3775, "total_steps": 6870, "loss": 0.2688, "lr": 2.505716138857851e-05, "epoch": 1.6484716157205241, "percentage": 54.95, "elapsed_time": "15:00:18", "remaining_time": "12:18:07"} +{"current_steps": 3780, "total_steps": 6870, "loss": 0.2738, "lr": 2.499364872913676e-05, "epoch": 1.6506550218340612, "percentage": 55.02, "elapsed_time": "15:01:29", "remaining_time": "12:16:56"} +{"current_steps": 3785, "total_steps": 6870, "loss": 0.2712, "lr": 2.4930136110687246e-05, "epoch": 1.6528384279475983, "percentage": 55.09, "elapsed_time": "15:02:41", "remaining_time": "12:15:44"} +{"current_steps": 3790, "total_steps": 6870, "loss": 0.2864, "lr": 2.486662394315181e-05, "epoch": 1.6550218340611353, "percentage": 55.17, "elapsed_time": "15:03:52", "remaining_time": "12:14:33"} +{"current_steps": 3795, "total_steps": 6870, "loss": 0.2732, "lr": 2.4803112636449355e-05, "epoch": 1.6572052401746724, "percentage": 55.24, "elapsed_time": "15:05:04", "remaining_time": "12:13:21"} +{"current_steps": 3800, "total_steps": 6870, "loss": 0.2789, "lr": 2.473960260049325e-05, "epoch": 1.6593886462882095, "percentage": 55.31, "elapsed_time": "15:06:16", "remaining_time": "12:12:10"} +{"current_steps": 3805, "total_steps": 6870, "loss": 0.2929, "lr": 2.4676094245188658e-05, "epoch": 1.6615720524017468, "percentage": 55.39, "elapsed_time": "15:07:27", "remaining_time": "12:10:58"} +{"current_steps": 3810, "total_steps": 6870, "loss": 0.2683, "lr": 2.461258798042989e-05, "epoch": 1.6637554585152838, "percentage": 55.46, "elapsed_time": "15:08:39", "remaining_time": "12:09:47"} +{"current_steps": 3815, "total_steps": 6870, "loss": 0.2552, "lr": 2.454908421609777e-05, "epoch": 1.6659388646288211, "percentage": 55.53, "elapsed_time": "15:09:50", "remaining_time": "12:08:35"} +{"current_steps": 3820, "total_steps": 6870, "loss": 0.2785, "lr": 2.4485583362056974e-05, "epoch": 1.6681222707423582, "percentage": 55.6, "elapsed_time": "15:11:02", "remaining_time": "12:07:24"} +{"current_steps": 3825, "total_steps": 6870, "loss": 0.2661, "lr": 2.4422085828153403e-05, "epoch": 1.6703056768558953, "percentage": 55.68, "elapsed_time": "15:12:13", "remaining_time": "12:06:12"} +{"current_steps": 3830, "total_steps": 6870, "loss": 0.2688, "lr": 2.435859202421153e-05, "epoch": 1.6724890829694323, "percentage": 55.75, "elapsed_time": "15:13:25", "remaining_time": "12:05:00"} +{"current_steps": 3835, "total_steps": 6870, "loss": 0.2602, "lr": 2.429510236003175e-05, "epoch": 1.6746724890829694, "percentage": 55.82, "elapsed_time": "15:14:37", "remaining_time": "12:03:49"} +{"current_steps": 3840, "total_steps": 6870, "loss": 0.2713, "lr": 2.4231617245387745e-05, "epoch": 1.6768558951965065, "percentage": 55.9, "elapsed_time": "15:15:48", "remaining_time": "12:02:37"} +{"current_steps": 3845, "total_steps": 6870, "loss": 0.2712, "lr": 2.416813709002382e-05, "epoch": 1.6790393013100435, "percentage": 55.97, "elapsed_time": "15:17:00", "remaining_time": "12:01:26"} +{"current_steps": 3850, "total_steps": 6870, "loss": 0.2863, "lr": 2.4104662303652288e-05, "epoch": 1.6812227074235808, "percentage": 56.04, "elapsed_time": "15:18:11", "remaining_time": "12:00:14"} +{"current_steps": 3855, "total_steps": 6870, "loss": 0.2679, "lr": 2.4041193295950788e-05, "epoch": 1.6834061135371179, "percentage": 56.11, "elapsed_time": "15:19:23", "remaining_time": "11:59:03"} +{"current_steps": 3860, "total_steps": 6870, "loss": 0.2762, "lr": 2.39777304765597e-05, "epoch": 1.6855895196506552, "percentage": 56.19, "elapsed_time": "15:20:34", "remaining_time": "11:57:51"} +{"current_steps": 3865, "total_steps": 6870, "loss": 0.2768, "lr": 2.391427425507943e-05, "epoch": 1.6877729257641922, "percentage": 56.26, "elapsed_time": "15:21:46", "remaining_time": "11:56:40"} +{"current_steps": 3870, "total_steps": 6870, "loss": 0.2659, "lr": 2.385082504106781e-05, "epoch": 1.6899563318777293, "percentage": 56.33, "elapsed_time": "15:22:58", "remaining_time": "11:55:28"} +{"current_steps": 3875, "total_steps": 6870, "loss": 0.2651, "lr": 2.378738324403744e-05, "epoch": 1.6921397379912664, "percentage": 56.4, "elapsed_time": "15:24:09", "remaining_time": "11:54:17"} +{"current_steps": 3880, "total_steps": 6870, "loss": 0.267, "lr": 2.3723949273453073e-05, "epoch": 1.6943231441048034, "percentage": 56.48, "elapsed_time": "15:25:21", "remaining_time": "11:53:05"} +{"current_steps": 3885, "total_steps": 6870, "loss": 0.2673, "lr": 2.3660523538728904e-05, "epoch": 1.6965065502183405, "percentage": 56.55, "elapsed_time": "15:26:32", "remaining_time": "11:51:54"} +{"current_steps": 3890, "total_steps": 6870, "loss": 0.2603, "lr": 2.3597106449226015e-05, "epoch": 1.6986899563318776, "percentage": 56.62, "elapsed_time": "15:27:44", "remaining_time": "11:50:42"} +{"current_steps": 3895, "total_steps": 6870, "loss": 0.2534, "lr": 2.3533698414249668e-05, "epoch": 1.7008733624454149, "percentage": 56.7, "elapsed_time": "15:28:55", "remaining_time": "11:49:31"} +{"current_steps": 3900, "total_steps": 6870, "loss": 0.2607, "lr": 2.3470299843046693e-05, "epoch": 1.703056768558952, "percentage": 56.77, "elapsed_time": "15:30:07", "remaining_time": "11:48:19"} +{"current_steps": 3905, "total_steps": 6870, "loss": 0.2674, "lr": 2.340691114480284e-05, "epoch": 1.7052401746724892, "percentage": 56.84, "elapsed_time": "15:31:19", "remaining_time": "11:47:08"} +{"current_steps": 3910, "total_steps": 6870, "loss": 0.2729, "lr": 2.334353272864013e-05, "epoch": 1.7074235807860263, "percentage": 56.91, "elapsed_time": "15:32:30", "remaining_time": "11:45:56"} +{"current_steps": 3915, "total_steps": 6870, "loss": 0.2665, "lr": 2.3280165003614227e-05, "epoch": 1.7096069868995634, "percentage": 56.99, "elapsed_time": "15:33:42", "remaining_time": "11:44:45"} +{"current_steps": 3920, "total_steps": 6870, "loss": 0.2707, "lr": 2.3216808378711788e-05, "epoch": 1.7117903930131004, "percentage": 57.06, "elapsed_time": "15:34:53", "remaining_time": "11:43:33"} +{"current_steps": 3925, "total_steps": 6870, "loss": 0.2418, "lr": 2.3153463262847835e-05, "epoch": 1.7139737991266375, "percentage": 57.13, "elapsed_time": "15:36:05", "remaining_time": "11:42:21"} +{"current_steps": 3930, "total_steps": 6870, "loss": 0.268, "lr": 2.3090130064863113e-05, "epoch": 1.7161572052401746, "percentage": 57.21, "elapsed_time": "15:37:16", "remaining_time": "11:41:10"} +{"current_steps": 3935, "total_steps": 6870, "loss": 0.2748, "lr": 2.3026809193521423e-05, "epoch": 1.7183406113537116, "percentage": 57.28, "elapsed_time": "15:38:28", "remaining_time": "11:39:58"} +{"current_steps": 3940, "total_steps": 6870, "loss": 0.2681, "lr": 2.296350105750703e-05, "epoch": 1.720524017467249, "percentage": 57.35, "elapsed_time": "15:39:40", "remaining_time": "11:38:47"} +{"current_steps": 3945, "total_steps": 6870, "loss": 0.2599, "lr": 2.2900206065421997e-05, "epoch": 1.722707423580786, "percentage": 57.42, "elapsed_time": "15:40:52", "remaining_time": "11:37:36"} +{"current_steps": 3950, "total_steps": 6870, "loss": 0.2775, "lr": 2.2836924625783554e-05, "epoch": 1.7248908296943233, "percentage": 57.5, "elapsed_time": "15:42:04", "remaining_time": "11:36:24"} +{"current_steps": 3955, "total_steps": 6870, "loss": 0.2811, "lr": 2.2773657147021464e-05, "epoch": 1.7270742358078603, "percentage": 57.57, "elapsed_time": "15:43:15", "remaining_time": "11:35:13"} +{"current_steps": 3960, "total_steps": 6870, "loss": 0.2596, "lr": 2.271040403747537e-05, "epoch": 1.7292576419213974, "percentage": 57.64, "elapsed_time": "15:44:27", "remaining_time": "11:34:01"} +{"current_steps": 3965, "total_steps": 6870, "loss": 0.2698, "lr": 2.2647165705392186e-05, "epoch": 1.7314410480349345, "percentage": 57.71, "elapsed_time": "15:45:38", "remaining_time": "11:32:50"} +{"current_steps": 3970, "total_steps": 6870, "loss": 0.272, "lr": 2.2583942558923456e-05, "epoch": 1.7336244541484715, "percentage": 57.79, "elapsed_time": "15:46:50", "remaining_time": "11:31:38"} +{"current_steps": 3975, "total_steps": 6870, "loss": 0.2745, "lr": 2.2520735006122688e-05, "epoch": 1.7358078602620086, "percentage": 57.86, "elapsed_time": "15:48:01", "remaining_time": "11:30:27"} +{"current_steps": 3980, "total_steps": 6870, "loss": 0.2588, "lr": 2.2457543454942786e-05, "epoch": 1.737991266375546, "percentage": 57.93, "elapsed_time": "15:49:13", "remaining_time": "11:29:15"} +{"current_steps": 3985, "total_steps": 6870, "loss": 0.2622, "lr": 2.2394368313233332e-05, "epoch": 1.740174672489083, "percentage": 58.01, "elapsed_time": "15:50:25", "remaining_time": "11:28:04"} +{"current_steps": 3990, "total_steps": 6870, "loss": 0.2603, "lr": 2.2331209988738024e-05, "epoch": 1.74235807860262, "percentage": 58.08, "elapsed_time": "15:51:36", "remaining_time": "11:26:52"} +{"current_steps": 3995, "total_steps": 6870, "loss": 0.2621, "lr": 2.226806888909202e-05, "epoch": 1.7445414847161573, "percentage": 58.15, "elapsed_time": "15:52:48", "remaining_time": "11:25:41"} +{"current_steps": 4000, "total_steps": 6870, "loss": 0.2636, "lr": 2.2204945421819292e-05, "epoch": 1.7467248908296944, "percentage": 58.22, "elapsed_time": "15:53:59", "remaining_time": "11:24:29"} +{"current_steps": 4005, "total_steps": 6870, "loss": 0.2744, "lr": 2.2141839994330027e-05, "epoch": 1.7489082969432315, "percentage": 58.3, "elapsed_time": "15:55:11", "remaining_time": "11:23:18"} +{"current_steps": 4010, "total_steps": 6870, "loss": 0.2711, "lr": 2.207875301391796e-05, "epoch": 1.7510917030567685, "percentage": 58.37, "elapsed_time": "15:56:22", "remaining_time": "11:22:06"} +{"current_steps": 4015, "total_steps": 6870, "loss": 0.2542, "lr": 2.2015684887757776e-05, "epoch": 1.7532751091703056, "percentage": 58.44, "elapsed_time": "15:57:34", "remaining_time": "11:20:54"} +{"current_steps": 4020, "total_steps": 6870, "loss": 0.2757, "lr": 2.195263602290248e-05, "epoch": 1.7554585152838427, "percentage": 58.52, "elapsed_time": "15:58:46", "remaining_time": "11:19:43"} +{"current_steps": 4025, "total_steps": 6870, "loss": 0.2469, "lr": 2.188960682628074e-05, "epoch": 1.75764192139738, "percentage": 58.59, "elapsed_time": "15:59:57", "remaining_time": "11:18:31"} +{"current_steps": 4030, "total_steps": 6870, "loss": 0.2707, "lr": 2.1826597704694306e-05, "epoch": 1.759825327510917, "percentage": 58.66, "elapsed_time": "16:01:09", "remaining_time": "11:17:20"} +{"current_steps": 4035, "total_steps": 6870, "loss": 0.2701, "lr": 2.176360906481533e-05, "epoch": 1.762008733624454, "percentage": 58.73, "elapsed_time": "16:02:20", "remaining_time": "11:16:08"} +{"current_steps": 4040, "total_steps": 6870, "loss": 0.2606, "lr": 2.1700641313183807e-05, "epoch": 1.7641921397379914, "percentage": 58.81, "elapsed_time": "16:03:32", "remaining_time": "11:14:57"} +{"current_steps": 4045, "total_steps": 6870, "loss": 0.2646, "lr": 2.163769485620489e-05, "epoch": 1.7663755458515285, "percentage": 58.88, "elapsed_time": "16:04:43", "remaining_time": "11:13:45"} +{"current_steps": 4050, "total_steps": 6870, "loss": 0.2721, "lr": 2.1574770100146295e-05, "epoch": 1.7685589519650655, "percentage": 58.95, "elapsed_time": "16:05:55", "remaining_time": "11:12:34"} +{"current_steps": 4055, "total_steps": 6870, "loss": 0.2598, "lr": 2.1511867451135693e-05, "epoch": 1.7707423580786026, "percentage": 59.02, "elapsed_time": "16:07:06", "remaining_time": "11:11:22"} +{"current_steps": 4060, "total_steps": 6870, "loss": 0.2687, "lr": 2.1448987315158048e-05, "epoch": 1.7729257641921397, "percentage": 59.1, "elapsed_time": "16:08:18", "remaining_time": "11:10:11"} +{"current_steps": 4065, "total_steps": 6870, "loss": 0.263, "lr": 2.1386130098053036e-05, "epoch": 1.7751091703056767, "percentage": 59.17, "elapsed_time": "16:09:30", "remaining_time": "11:08:59"} +{"current_steps": 4070, "total_steps": 6870, "loss": 0.271, "lr": 2.1323296205512418e-05, "epoch": 1.777292576419214, "percentage": 59.24, "elapsed_time": "16:10:41", "remaining_time": "11:07:47"} +{"current_steps": 4075, "total_steps": 6870, "loss": 0.258, "lr": 2.1260486043077388e-05, "epoch": 1.779475982532751, "percentage": 59.32, "elapsed_time": "16:11:53", "remaining_time": "11:06:36"} +{"current_steps": 4080, "total_steps": 6870, "loss": 0.2662, "lr": 2.1197700016136014e-05, "epoch": 1.7816593886462884, "percentage": 59.39, "elapsed_time": "16:13:04", "remaining_time": "11:05:24"} +{"current_steps": 4085, "total_steps": 6870, "loss": 0.2574, "lr": 2.113493852992056e-05, "epoch": 1.7838427947598254, "percentage": 59.46, "elapsed_time": "16:14:16", "remaining_time": "11:04:13"} +{"current_steps": 4090, "total_steps": 6870, "loss": 0.273, "lr": 2.1072201989504913e-05, "epoch": 1.7860262008733625, "percentage": 59.53, "elapsed_time": "16:15:27", "remaining_time": "11:03:01"} +{"current_steps": 4095, "total_steps": 6870, "loss": 0.2692, "lr": 2.1009490799801965e-05, "epoch": 1.7882096069868996, "percentage": 59.61, "elapsed_time": "16:16:39", "remaining_time": "11:01:50"} +{"current_steps": 4100, "total_steps": 6870, "loss": 0.2745, "lr": 2.0946805365560966e-05, "epoch": 1.7903930131004366, "percentage": 59.68, "elapsed_time": "16:17:50", "remaining_time": "11:00:38"} +{"current_steps": 4105, "total_steps": 6870, "loss": 0.2659, "lr": 2.0884146091364956e-05, "epoch": 1.7925764192139737, "percentage": 59.75, "elapsed_time": "16:19:02", "remaining_time": "10:59:27"} +{"current_steps": 4110, "total_steps": 6870, "loss": 0.2648, "lr": 2.0821513381628132e-05, "epoch": 1.7947598253275108, "percentage": 59.83, "elapsed_time": "16:20:13", "remaining_time": "10:58:15"} +{"current_steps": 4115, "total_steps": 6870, "loss": 0.2564, "lr": 2.0758907640593228e-05, "epoch": 1.796943231441048, "percentage": 59.9, "elapsed_time": "16:21:25", "remaining_time": "10:57:03"} +{"current_steps": 4120, "total_steps": 6870, "loss": 0.2723, "lr": 2.0696329272328932e-05, "epoch": 1.7991266375545851, "percentage": 59.97, "elapsed_time": "16:22:36", "remaining_time": "10:55:52"} +{"current_steps": 4125, "total_steps": 6870, "loss": 0.2616, "lr": 2.063377868072724e-05, "epoch": 1.8013100436681224, "percentage": 60.04, "elapsed_time": "16:23:48", "remaining_time": "10:54:40"} +{"current_steps": 4130, "total_steps": 6870, "loss": 0.27, "lr": 2.0571256269500893e-05, "epoch": 1.8034934497816595, "percentage": 60.12, "elapsed_time": "16:25:00", "remaining_time": "10:53:29"} +{"current_steps": 4135, "total_steps": 6870, "loss": 0.2678, "lr": 2.0508762442180745e-05, "epoch": 1.8056768558951966, "percentage": 60.19, "elapsed_time": "16:26:11", "remaining_time": "10:52:17"} +{"current_steps": 4140, "total_steps": 6870, "loss": 0.2777, "lr": 2.0446297602113167e-05, "epoch": 1.8078602620087336, "percentage": 60.26, "elapsed_time": "16:27:23", "remaining_time": "10:51:06"} +{"current_steps": 4145, "total_steps": 6870, "loss": 0.2632, "lr": 2.0383862152457432e-05, "epoch": 1.8100436681222707, "percentage": 60.33, "elapsed_time": "16:28:34", "remaining_time": "10:49:54"} +{"current_steps": 4150, "total_steps": 6870, "loss": 0.2756, "lr": 2.0321456496183134e-05, "epoch": 1.8122270742358078, "percentage": 60.41, "elapsed_time": "16:29:46", "remaining_time": "10:48:43"} +{"current_steps": 4155, "total_steps": 6870, "loss": 0.2703, "lr": 2.0259081036067563e-05, "epoch": 1.8144104803493448, "percentage": 60.48, "elapsed_time": "16:30:57", "remaining_time": "10:47:31"} +{"current_steps": 4160, "total_steps": 6870, "loss": 0.2527, "lr": 2.0196736174693126e-05, "epoch": 1.8165938864628821, "percentage": 60.55, "elapsed_time": "16:32:09", "remaining_time": "10:46:19"} +{"current_steps": 4165, "total_steps": 6870, "loss": 0.2642, "lr": 2.013442231444474e-05, "epoch": 1.8187772925764192, "percentage": 60.63, "elapsed_time": "16:33:20", "remaining_time": "10:45:08"} +{"current_steps": 4170, "total_steps": 6870, "loss": 0.2651, "lr": 2.007213985750724e-05, "epoch": 1.8209606986899565, "percentage": 60.7, "elapsed_time": "16:34:32", "remaining_time": "10:43:56"} +{"current_steps": 4175, "total_steps": 6870, "loss": 0.2653, "lr": 2.000988920586275e-05, "epoch": 1.8231441048034935, "percentage": 60.77, "elapsed_time": "16:35:43", "remaining_time": "10:42:45"} +{"current_steps": 4180, "total_steps": 6870, "loss": 0.2571, "lr": 1.994767076128816e-05, "epoch": 1.8253275109170306, "percentage": 60.84, "elapsed_time": "16:36:55", "remaining_time": "10:41:33"} +{"current_steps": 4185, "total_steps": 6870, "loss": 0.2754, "lr": 1.988548492535246e-05, "epoch": 1.8275109170305677, "percentage": 60.92, "elapsed_time": "16:38:06", "remaining_time": "10:40:22"} +{"current_steps": 4190, "total_steps": 6870, "loss": 0.2717, "lr": 1.9823332099414185e-05, "epoch": 1.8296943231441047, "percentage": 60.99, "elapsed_time": "16:39:18", "remaining_time": "10:39:10"} +{"current_steps": 4195, "total_steps": 6870, "loss": 0.2502, "lr": 1.9761212684618824e-05, "epoch": 1.8318777292576418, "percentage": 61.06, "elapsed_time": "16:40:30", "remaining_time": "10:37:58"} +{"current_steps": 4200, "total_steps": 6870, "loss": 0.2514, "lr": 1.9699127081896214e-05, "epoch": 1.8340611353711789, "percentage": 61.14, "elapsed_time": "16:41:41", "remaining_time": "10:36:47"} +{"current_steps": 4205, "total_steps": 6870, "loss": 0.2523, "lr": 1.963707569195797e-05, "epoch": 1.8362445414847162, "percentage": 61.21, "elapsed_time": "16:42:53", "remaining_time": "10:35:35"} +{"current_steps": 4210, "total_steps": 6870, "loss": 0.2649, "lr": 1.9575058915294885e-05, "epoch": 1.8384279475982532, "percentage": 61.28, "elapsed_time": "16:44:04", "remaining_time": "10:34:24"} +{"current_steps": 4215, "total_steps": 6870, "loss": 0.2778, "lr": 1.9513077152174363e-05, "epoch": 1.8406113537117905, "percentage": 61.35, "elapsed_time": "16:45:16", "remaining_time": "10:33:12"} +{"current_steps": 4220, "total_steps": 6870, "loss": 0.2588, "lr": 1.9451130802637822e-05, "epoch": 1.8427947598253276, "percentage": 61.43, "elapsed_time": "16:46:27", "remaining_time": "10:32:01"} +{"current_steps": 4225, "total_steps": 6870, "loss": 0.2701, "lr": 1.9389220266498102e-05, "epoch": 1.8449781659388647, "percentage": 61.5, "elapsed_time": "16:47:39", "remaining_time": "10:30:49"} +{"current_steps": 4230, "total_steps": 6870, "loss": 0.2644, "lr": 1.9327345943336904e-05, "epoch": 1.8471615720524017, "percentage": 61.57, "elapsed_time": "16:48:50", "remaining_time": "10:29:38"} +{"current_steps": 4235, "total_steps": 6870, "loss": 0.2748, "lr": 1.9265508232502208e-05, "epoch": 1.8493449781659388, "percentage": 61.64, "elapsed_time": "16:50:02", "remaining_time": "10:28:26"} +{"current_steps": 4240, "total_steps": 6870, "loss": 0.26, "lr": 1.9203707533105687e-05, "epoch": 1.8515283842794759, "percentage": 61.72, "elapsed_time": "16:51:14", "remaining_time": "10:27:15"} +{"current_steps": 4245, "total_steps": 6870, "loss": 0.2711, "lr": 1.914194424402015e-05, "epoch": 1.8537117903930131, "percentage": 61.79, "elapsed_time": "16:52:25", "remaining_time": "10:26:03"} +{"current_steps": 4250, "total_steps": 6870, "loss": 0.2512, "lr": 1.9080218763876917e-05, "epoch": 1.8558951965065502, "percentage": 61.86, "elapsed_time": "16:53:37", "remaining_time": "10:24:51"} +{"current_steps": 4255, "total_steps": 6870, "loss": 0.2627, "lr": 1.901853149106332e-05, "epoch": 1.8580786026200873, "percentage": 61.94, "elapsed_time": "16:54:48", "remaining_time": "10:23:40"} +{"current_steps": 4260, "total_steps": 6870, "loss": 0.2657, "lr": 1.8956882823720075e-05, "epoch": 1.8602620087336246, "percentage": 62.01, "elapsed_time": "16:56:00", "remaining_time": "10:22:28"} +{"current_steps": 4265, "total_steps": 6870, "loss": 0.2599, "lr": 1.889527315973874e-05, "epoch": 1.8624454148471616, "percentage": 62.08, "elapsed_time": "16:57:11", "remaining_time": "10:21:17"} +{"current_steps": 4270, "total_steps": 6870, "loss": 0.2636, "lr": 1.8833702896759132e-05, "epoch": 1.8646288209606987, "percentage": 62.15, "elapsed_time": "16:58:23", "remaining_time": "10:20:05"} +{"current_steps": 4275, "total_steps": 6870, "loss": 0.2744, "lr": 1.8772172432166767e-05, "epoch": 1.8668122270742358, "percentage": 62.23, "elapsed_time": "16:59:35", "remaining_time": "10:18:54"} +{"current_steps": 4280, "total_steps": 6870, "loss": 0.2744, "lr": 1.87106821630903e-05, "epoch": 1.8689956331877728, "percentage": 62.3, "elapsed_time": "17:00:46", "remaining_time": "10:17:42"} +{"current_steps": 4285, "total_steps": 6870, "loss": 0.2637, "lr": 1.8649232486398955e-05, "epoch": 1.87117903930131, "percentage": 62.37, "elapsed_time": "17:01:58", "remaining_time": "10:16:31"} +{"current_steps": 4290, "total_steps": 6870, "loss": 0.2598, "lr": 1.8587823798699958e-05, "epoch": 1.8733624454148472, "percentage": 62.45, "elapsed_time": "17:03:09", "remaining_time": "10:15:19"} +{"current_steps": 4295, "total_steps": 6870, "loss": 0.2528, "lr": 1.8526456496335997e-05, "epoch": 1.8755458515283843, "percentage": 62.52, "elapsed_time": "17:04:21", "remaining_time": "10:14:08"} +{"current_steps": 4300, "total_steps": 6870, "loss": 0.2548, "lr": 1.846513097538265e-05, "epoch": 1.8777292576419216, "percentage": 62.59, "elapsed_time": "17:05:32", "remaining_time": "10:12:56"} +{"current_steps": 4305, "total_steps": 6870, "loss": 0.2747, "lr": 1.8403847631645815e-05, "epoch": 1.8799126637554586, "percentage": 62.66, "elapsed_time": "17:06:44", "remaining_time": "10:11:45"} +{"current_steps": 4310, "total_steps": 6870, "loss": 0.2754, "lr": 1.834260686065919e-05, "epoch": 1.8820960698689957, "percentage": 62.74, "elapsed_time": "17:07:56", "remaining_time": "10:10:33"} +{"current_steps": 4315, "total_steps": 6870, "loss": 0.2556, "lr": 1.8281409057681685e-05, "epoch": 1.8842794759825328, "percentage": 62.81, "elapsed_time": "17:09:07", "remaining_time": "10:09:22"} +{"current_steps": 4320, "total_steps": 6870, "loss": 0.2728, "lr": 1.8220254617694904e-05, "epoch": 1.8864628820960698, "percentage": 62.88, "elapsed_time": "17:10:19", "remaining_time": "10:08:10"} +{"current_steps": 4325, "total_steps": 6870, "loss": 0.2674, "lr": 1.815914393540057e-05, "epoch": 1.888646288209607, "percentage": 62.95, "elapsed_time": "17:11:30", "remaining_time": "10:06:58"} +{"current_steps": 4330, "total_steps": 6870, "loss": 0.2682, "lr": 1.8098077405217973e-05, "epoch": 1.890829694323144, "percentage": 63.03, "elapsed_time": "17:12:42", "remaining_time": "10:05:47"} +{"current_steps": 4335, "total_steps": 6870, "loss": 0.2755, "lr": 1.8037055421281453e-05, "epoch": 1.8930131004366813, "percentage": 63.1, "elapsed_time": "17:13:53", "remaining_time": "10:04:35"} +{"current_steps": 4340, "total_steps": 6870, "loss": 0.2538, "lr": 1.7976078377437843e-05, "epoch": 1.8951965065502183, "percentage": 63.17, "elapsed_time": "17:15:05", "remaining_time": "10:03:24"} +{"current_steps": 4345, "total_steps": 6870, "loss": 0.2707, "lr": 1.7915146667243912e-05, "epoch": 1.8973799126637556, "percentage": 63.25, "elapsed_time": "17:16:16", "remaining_time": "10:02:12"} +{"current_steps": 4350, "total_steps": 6870, "loss": 0.2724, "lr": 1.785426068396385e-05, "epoch": 1.8995633187772927, "percentage": 63.32, "elapsed_time": "17:17:28", "remaining_time": "10:01:01"} +{"current_steps": 4355, "total_steps": 6870, "loss": 0.3122, "lr": 1.7793420820566698e-05, "epoch": 1.9017467248908297, "percentage": 63.39, "elapsed_time": "17:18:40", "remaining_time": "9:59:49"} +{"current_steps": 4360, "total_steps": 6870, "loss": 0.2669, "lr": 1.7732627469723867e-05, "epoch": 1.9039301310043668, "percentage": 63.46, "elapsed_time": "17:19:51", "remaining_time": "9:58:38"} +{"current_steps": 4365, "total_steps": 6870, "loss": 0.2633, "lr": 1.7671881023806526e-05, "epoch": 1.9061135371179039, "percentage": 63.54, "elapsed_time": "17:21:03", "remaining_time": "9:57:26"} +{"current_steps": 4370, "total_steps": 6870, "loss": 0.264, "lr": 1.7611181874883133e-05, "epoch": 1.908296943231441, "percentage": 63.61, "elapsed_time": "17:22:14", "remaining_time": "9:56:15"} +{"current_steps": 4375, "total_steps": 6870, "loss": 0.2631, "lr": 1.755053041471689e-05, "epoch": 1.910480349344978, "percentage": 63.68, "elapsed_time": "17:23:26", "remaining_time": "9:55:03"} +{"current_steps": 4380, "total_steps": 6870, "loss": 0.2624, "lr": 1.748992703476318e-05, "epoch": 1.9126637554585153, "percentage": 63.76, "elapsed_time": "17:24:37", "remaining_time": "9:53:51"} +{"current_steps": 4385, "total_steps": 6870, "loss": 0.2493, "lr": 1.7429372126167105e-05, "epoch": 1.9148471615720524, "percentage": 63.83, "elapsed_time": "17:25:49", "remaining_time": "9:52:40"} +{"current_steps": 4390, "total_steps": 6870, "loss": 0.2617, "lr": 1.7368866079760887e-05, "epoch": 1.9170305676855897, "percentage": 63.9, "elapsed_time": "17:27:01", "remaining_time": "9:51:28"} +{"current_steps": 4395, "total_steps": 6870, "loss": 0.2688, "lr": 1.7308409286061407e-05, "epoch": 1.9192139737991267, "percentage": 63.97, "elapsed_time": "17:28:12", "remaining_time": "9:50:17"} +{"current_steps": 4400, "total_steps": 6870, "loss": 0.2668, "lr": 1.7248002135267645e-05, "epoch": 1.9213973799126638, "percentage": 64.05, "elapsed_time": "17:29:24", "remaining_time": "9:49:05"} +{"current_steps": 4405, "total_steps": 6870, "loss": 0.2736, "lr": 1.7187645017258196e-05, "epoch": 1.9235807860262009, "percentage": 64.12, "elapsed_time": "17:30:35", "remaining_time": "9:47:54"} +{"current_steps": 4410, "total_steps": 6870, "loss": 0.2618, "lr": 1.712733832158872e-05, "epoch": 1.925764192139738, "percentage": 64.19, "elapsed_time": "17:31:47", "remaining_time": "9:46:42"} +{"current_steps": 4415, "total_steps": 6870, "loss": 0.2656, "lr": 1.7067082437489436e-05, "epoch": 1.927947598253275, "percentage": 64.26, "elapsed_time": "17:32:58", "remaining_time": "9:45:31"} +{"current_steps": 4420, "total_steps": 6870, "loss": 0.2625, "lr": 1.7006877753862632e-05, "epoch": 1.930131004366812, "percentage": 64.34, "elapsed_time": "17:34:10", "remaining_time": "9:44:19"} +{"current_steps": 4425, "total_steps": 6870, "loss": 0.2544, "lr": 1.6946724659280133e-05, "epoch": 1.9323144104803494, "percentage": 64.41, "elapsed_time": "17:35:21", "remaining_time": "9:43:07"} +{"current_steps": 4430, "total_steps": 6870, "loss": 0.2759, "lr": 1.6886623541980796e-05, "epoch": 1.9344978165938864, "percentage": 64.48, "elapsed_time": "17:36:33", "remaining_time": "9:41:56"} +{"current_steps": 4435, "total_steps": 6870, "loss": 0.2646, "lr": 1.682657478986801e-05, "epoch": 1.9366812227074237, "percentage": 64.56, "elapsed_time": "17:37:44", "remaining_time": "9:40:44"} +{"current_steps": 4440, "total_steps": 6870, "loss": 0.2597, "lr": 1.6766578790507192e-05, "epoch": 1.9388646288209608, "percentage": 64.63, "elapsed_time": "17:38:56", "remaining_time": "9:39:33"} +{"current_steps": 4445, "total_steps": 6870, "loss": 0.26, "lr": 1.6706635931123277e-05, "epoch": 1.9410480349344978, "percentage": 64.7, "elapsed_time": "17:40:07", "remaining_time": "9:38:21"} +{"current_steps": 4450, "total_steps": 6870, "loss": 0.2493, "lr": 1.664674659859822e-05, "epoch": 1.943231441048035, "percentage": 64.77, "elapsed_time": "17:41:19", "remaining_time": "9:37:10"} +{"current_steps": 4455, "total_steps": 6870, "loss": 0.2762, "lr": 1.658691117946853e-05, "epoch": 1.945414847161572, "percentage": 64.85, "elapsed_time": "17:42:31", "remaining_time": "9:35:58"} +{"current_steps": 4460, "total_steps": 6870, "loss": 0.2634, "lr": 1.6527130059922717e-05, "epoch": 1.947598253275109, "percentage": 64.92, "elapsed_time": "17:43:42", "remaining_time": "9:34:47"} +{"current_steps": 4465, "total_steps": 6870, "loss": 0.2617, "lr": 1.646740362579886e-05, "epoch": 1.9497816593886463, "percentage": 64.99, "elapsed_time": "17:44:54", "remaining_time": "9:33:35"} +{"current_steps": 4470, "total_steps": 6870, "loss": 0.2658, "lr": 1.6407732262582067e-05, "epoch": 1.9519650655021834, "percentage": 65.07, "elapsed_time": "17:46:05", "remaining_time": "9:32:23"} +{"current_steps": 4475, "total_steps": 6870, "loss": 0.2594, "lr": 1.634811635540201e-05, "epoch": 1.9541484716157205, "percentage": 65.14, "elapsed_time": "17:47:17", "remaining_time": "9:31:12"} +{"current_steps": 4480, "total_steps": 6870, "loss": 0.2678, "lr": 1.6288556289030452e-05, "epoch": 1.9563318777292578, "percentage": 65.21, "elapsed_time": "17:48:28", "remaining_time": "9:30:00"} +{"current_steps": 4485, "total_steps": 6870, "loss": 0.2543, "lr": 1.6229052447878743e-05, "epoch": 1.9585152838427948, "percentage": 65.28, "elapsed_time": "17:49:40", "remaining_time": "9:28:49"} +{"current_steps": 4490, "total_steps": 6870, "loss": 0.2589, "lr": 1.6169605215995345e-05, "epoch": 1.960698689956332, "percentage": 65.36, "elapsed_time": "17:50:51", "remaining_time": "9:27:37"} +{"current_steps": 4495, "total_steps": 6870, "loss": 0.256, "lr": 1.6110214977063345e-05, "epoch": 1.962882096069869, "percentage": 65.43, "elapsed_time": "17:52:03", "remaining_time": "9:26:26"} +{"current_steps": 4500, "total_steps": 6870, "loss": 0.2508, "lr": 1.6050882114398e-05, "epoch": 1.965065502183406, "percentage": 65.5, "elapsed_time": "17:53:14", "remaining_time": "9:25:14"} +{"current_steps": 4505, "total_steps": 6870, "loss": 0.2609, "lr": 1.5991607010944242e-05, "epoch": 1.967248908296943, "percentage": 65.57, "elapsed_time": "17:54:26", "remaining_time": "9:24:03"} +{"current_steps": 4510, "total_steps": 6870, "loss": 0.2642, "lr": 1.5932390049274227e-05, "epoch": 1.9694323144104804, "percentage": 65.65, "elapsed_time": "17:55:38", "remaining_time": "9:22:51"} +{"current_steps": 4515, "total_steps": 6870, "loss": 0.2722, "lr": 1.587323161158484e-05, "epoch": 1.9716157205240175, "percentage": 65.72, "elapsed_time": "17:56:49", "remaining_time": "9:21:40"} +{"current_steps": 4520, "total_steps": 6870, "loss": 0.2651, "lr": 1.5814132079695232e-05, "epoch": 1.9737991266375547, "percentage": 65.79, "elapsed_time": "17:58:01", "remaining_time": "9:20:28"} +{"current_steps": 4525, "total_steps": 6870, "loss": 0.2586, "lr": 1.57550918350444e-05, "epoch": 1.9759825327510918, "percentage": 65.87, "elapsed_time": "17:59:12", "remaining_time": "9:19:16"} +{"current_steps": 4530, "total_steps": 6870, "loss": 0.2556, "lr": 1.5696111258688657e-05, "epoch": 1.9781659388646289, "percentage": 65.94, "elapsed_time": "18:00:24", "remaining_time": "9:18:05"} +{"current_steps": 4535, "total_steps": 6870, "loss": 0.2656, "lr": 1.563719073129923e-05, "epoch": 1.980349344978166, "percentage": 66.01, "elapsed_time": "18:01:35", "remaining_time": "9:16:53"} +{"current_steps": 4540, "total_steps": 6870, "loss": 0.2594, "lr": 1.557833063315976e-05, "epoch": 1.982532751091703, "percentage": 66.08, "elapsed_time": "18:02:47", "remaining_time": "9:15:42"} +{"current_steps": 4545, "total_steps": 6870, "loss": 0.2596, "lr": 1.5519531344163873e-05, "epoch": 1.98471615720524, "percentage": 66.16, "elapsed_time": "18:03:59", "remaining_time": "9:14:30"} +{"current_steps": 4550, "total_steps": 6870, "loss": 0.2647, "lr": 1.5460793243812732e-05, "epoch": 1.9868995633187772, "percentage": 66.23, "elapsed_time": "18:05:10", "remaining_time": "9:13:19"} +{"current_steps": 4555, "total_steps": 6870, "loss": 0.2715, "lr": 1.5402116711212566e-05, "epoch": 1.9890829694323144, "percentage": 66.3, "elapsed_time": "18:06:22", "remaining_time": "9:12:07"} +{"current_steps": 4560, "total_steps": 6870, "loss": 0.2544, "lr": 1.5343502125072236e-05, "epoch": 1.9912663755458515, "percentage": 66.38, "elapsed_time": "18:07:33", "remaining_time": "9:10:56"} +{"current_steps": 4565, "total_steps": 6870, "loss": 0.2662, "lr": 1.5284949863700805e-05, "epoch": 1.9934497816593888, "percentage": 66.45, "elapsed_time": "18:08:45", "remaining_time": "9:09:44"} +{"current_steps": 4570, "total_steps": 6870, "loss": 0.2621, "lr": 1.5226460305005044e-05, "epoch": 1.9956331877729259, "percentage": 66.52, "elapsed_time": "18:09:56", "remaining_time": "9:08:32"} +{"current_steps": 4575, "total_steps": 6870, "loss": 0.2695, "lr": 1.5168033826487072e-05, "epoch": 1.997816593886463, "percentage": 66.59, "elapsed_time": "18:11:08", "remaining_time": "9:07:21"} +{"current_steps": 4580, "total_steps": 6870, "loss": 0.257, "lr": 1.5109670805241846e-05, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "18:12:19", "remaining_time": "9:06:09"} +{"current_steps": 4585, "total_steps": 6870, "loss": 0.2215, "lr": 1.505137161795478e-05, "epoch": 2.002183406113537, "percentage": 66.74, "elapsed_time": "18:13:31", "remaining_time": "9:04:58"} +{"current_steps": 4590, "total_steps": 6870, "loss": 0.2148, "lr": 1.4993136640899286e-05, "epoch": 2.004366812227074, "percentage": 66.81, "elapsed_time": "18:14:42", "remaining_time": "9:03:46"} +{"current_steps": 4595, "total_steps": 6870, "loss": 0.2053, "lr": 1.4934966249934334e-05, "epoch": 2.006550218340611, "percentage": 66.89, "elapsed_time": "18:15:54", "remaining_time": "9:02:35"} +{"current_steps": 4600, "total_steps": 6870, "loss": 0.2163, "lr": 1.487686082050207e-05, "epoch": 2.0087336244541483, "percentage": 66.96, "elapsed_time": "18:17:05", "remaining_time": "9:01:23"} +{"current_steps": 4605, "total_steps": 6870, "loss": 0.2071, "lr": 1.4818820727625348e-05, "epoch": 2.010917030567686, "percentage": 67.03, "elapsed_time": "18:18:17", "remaining_time": "9:00:12"} +{"current_steps": 4610, "total_steps": 6870, "loss": 0.2145, "lr": 1.4760846345905344e-05, "epoch": 2.013100436681223, "percentage": 67.1, "elapsed_time": "18:19:29", "remaining_time": "8:59:00"} +{"current_steps": 4615, "total_steps": 6870, "loss": 0.2168, "lr": 1.4702938049519105e-05, "epoch": 2.01528384279476, "percentage": 67.18, "elapsed_time": "18:20:40", "remaining_time": "8:57:49"} +{"current_steps": 4620, "total_steps": 6870, "loss": 0.2076, "lr": 1.4645096212217167e-05, "epoch": 2.017467248908297, "percentage": 67.25, "elapsed_time": "18:21:52", "remaining_time": "8:56:37"} +{"current_steps": 4625, "total_steps": 6870, "loss": 0.2233, "lr": 1.4587321207321115e-05, "epoch": 2.019650655021834, "percentage": 67.32, "elapsed_time": "18:23:03", "remaining_time": "8:55:26"} +{"current_steps": 4630, "total_steps": 6870, "loss": 0.2016, "lr": 1.4529613407721192e-05, "epoch": 2.021834061135371, "percentage": 67.39, "elapsed_time": "18:24:15", "remaining_time": "8:54:14"} +{"current_steps": 4635, "total_steps": 6870, "loss": 0.2091, "lr": 1.4471973185873888e-05, "epoch": 2.024017467248908, "percentage": 67.47, "elapsed_time": "18:25:27", "remaining_time": "8:53:02"} +{"current_steps": 4640, "total_steps": 6870, "loss": 0.2028, "lr": 1.4414400913799535e-05, "epoch": 2.0262008733624453, "percentage": 67.54, "elapsed_time": "18:26:38", "remaining_time": "8:51:51"} +{"current_steps": 4645, "total_steps": 6870, "loss": 0.2035, "lr": 1.4356896963079885e-05, "epoch": 2.0283842794759823, "percentage": 67.61, "elapsed_time": "18:27:50", "remaining_time": "8:50:39"} +{"current_steps": 4650, "total_steps": 6870, "loss": 0.2042, "lr": 1.4299461704855759e-05, "epoch": 2.03056768558952, "percentage": 67.69, "elapsed_time": "18:29:01", "remaining_time": "8:49:28"} +{"current_steps": 4655, "total_steps": 6870, "loss": 0.2151, "lr": 1.424209550982461e-05, "epoch": 2.032751091703057, "percentage": 67.76, "elapsed_time": "18:30:13", "remaining_time": "8:48:16"} +{"current_steps": 4660, "total_steps": 6870, "loss": 0.2175, "lr": 1.4184798748238148e-05, "epoch": 2.034934497816594, "percentage": 67.83, "elapsed_time": "18:31:24", "remaining_time": "8:47:05"} +{"current_steps": 4665, "total_steps": 6870, "loss": 0.2068, "lr": 1.4127571789899946e-05, "epoch": 2.037117903930131, "percentage": 67.9, "elapsed_time": "18:32:36", "remaining_time": "8:45:53"} +{"current_steps": 4670, "total_steps": 6870, "loss": 0.2164, "lr": 1.4070415004163056e-05, "epoch": 2.039301310043668, "percentage": 67.98, "elapsed_time": "18:33:48", "remaining_time": "8:44:42"} +{"current_steps": 4675, "total_steps": 6870, "loss": 0.2135, "lr": 1.4013328759927625e-05, "epoch": 2.041484716157205, "percentage": 68.05, "elapsed_time": "18:34:59", "remaining_time": "8:43:30"} +{"current_steps": 4680, "total_steps": 6870, "loss": 0.2017, "lr": 1.39563134256385e-05, "epoch": 2.0436681222707422, "percentage": 68.12, "elapsed_time": "18:36:11", "remaining_time": "8:42:19"} +{"current_steps": 4685, "total_steps": 6870, "loss": 0.2185, "lr": 1.3899369369282888e-05, "epoch": 2.0458515283842793, "percentage": 68.2, "elapsed_time": "18:37:22", "remaining_time": "8:41:07"} +{"current_steps": 4690, "total_steps": 6870, "loss": 0.2189, "lr": 1.3842496958387938e-05, "epoch": 2.0480349344978164, "percentage": 68.27, "elapsed_time": "18:38:34", "remaining_time": "8:39:56"} +{"current_steps": 4695, "total_steps": 6870, "loss": 0.204, "lr": 1.3785696560018374e-05, "epoch": 2.050218340611354, "percentage": 68.34, "elapsed_time": "18:39:45", "remaining_time": "8:38:44"} +{"current_steps": 4700, "total_steps": 6870, "loss": 0.206, "lr": 1.372896854077416e-05, "epoch": 2.052401746724891, "percentage": 68.41, "elapsed_time": "18:40:57", "remaining_time": "8:37:32"} +{"current_steps": 4705, "total_steps": 6870, "loss": 0.213, "lr": 1.3672313266788106e-05, "epoch": 2.054585152838428, "percentage": 68.49, "elapsed_time": "18:42:09", "remaining_time": "8:36:21"} +{"current_steps": 4710, "total_steps": 6870, "loss": 0.2125, "lr": 1.3615731103723507e-05, "epoch": 2.056768558951965, "percentage": 68.56, "elapsed_time": "18:43:20", "remaining_time": "8:35:09"} +{"current_steps": 4715, "total_steps": 6870, "loss": 0.2063, "lr": 1.3559222416771793e-05, "epoch": 2.058951965065502, "percentage": 68.63, "elapsed_time": "18:44:32", "remaining_time": "8:33:58"} +{"current_steps": 4720, "total_steps": 6870, "loss": 0.2108, "lr": 1.350278757065016e-05, "epoch": 2.061135371179039, "percentage": 68.7, "elapsed_time": "18:45:43", "remaining_time": "8:32:46"} +{"current_steps": 4725, "total_steps": 6870, "loss": 0.2139, "lr": 1.3446426929599229e-05, "epoch": 2.0633187772925763, "percentage": 68.78, "elapsed_time": "18:46:55", "remaining_time": "8:31:35"} +{"current_steps": 4730, "total_steps": 6870, "loss": 0.1966, "lr": 1.3390140857380678e-05, "epoch": 2.0655021834061134, "percentage": 68.85, "elapsed_time": "18:48:06", "remaining_time": "8:30:23"} +{"current_steps": 4735, "total_steps": 6870, "loss": 0.2256, "lr": 1.3333929717274919e-05, "epoch": 2.067685589519651, "percentage": 68.92, "elapsed_time": "18:49:18", "remaining_time": "8:29:12"} +{"current_steps": 4740, "total_steps": 6870, "loss": 0.2199, "lr": 1.3277793872078732e-05, "epoch": 2.069868995633188, "percentage": 69.0, "elapsed_time": "18:50:30", "remaining_time": "8:28:00"} +{"current_steps": 4745, "total_steps": 6870, "loss": 0.2129, "lr": 1.3221733684102905e-05, "epoch": 2.072052401746725, "percentage": 69.07, "elapsed_time": "18:51:41", "remaining_time": "8:26:49"} +{"current_steps": 4750, "total_steps": 6870, "loss": 0.2108, "lr": 1.3165749515169967e-05, "epoch": 2.074235807860262, "percentage": 69.14, "elapsed_time": "18:52:53", "remaining_time": "8:25:37"} +{"current_steps": 4755, "total_steps": 6870, "loss": 0.2182, "lr": 1.310984172661177e-05, "epoch": 2.076419213973799, "percentage": 69.21, "elapsed_time": "18:54:04", "remaining_time": "8:24:25"} +{"current_steps": 4760, "total_steps": 6870, "loss": 0.2145, "lr": 1.305401067926722e-05, "epoch": 2.078602620087336, "percentage": 69.29, "elapsed_time": "18:55:16", "remaining_time": "8:23:14"} +{"current_steps": 4765, "total_steps": 6870, "loss": 0.2003, "lr": 1.2998256733479897e-05, "epoch": 2.0807860262008733, "percentage": 69.36, "elapsed_time": "18:56:27", "remaining_time": "8:22:02"} +{"current_steps": 4770, "total_steps": 6870, "loss": 0.2042, "lr": 1.294258024909577e-05, "epoch": 2.0829694323144103, "percentage": 69.43, "elapsed_time": "18:57:39", "remaining_time": "8:20:51"} +{"current_steps": 4775, "total_steps": 6870, "loss": 0.2209, "lr": 1.288698158546085e-05, "epoch": 2.0851528384279474, "percentage": 69.51, "elapsed_time": "18:58:50", "remaining_time": "8:19:39"} +{"current_steps": 4780, "total_steps": 6870, "loss": 0.2172, "lr": 1.2831461101418884e-05, "epoch": 2.087336244541485, "percentage": 69.58, "elapsed_time": "19:00:02", "remaining_time": "8:18:28"} +{"current_steps": 4785, "total_steps": 6870, "loss": 0.2232, "lr": 1.2776019155309035e-05, "epoch": 2.089519650655022, "percentage": 69.65, "elapsed_time": "19:01:13", "remaining_time": "8:17:16"} +{"current_steps": 4790, "total_steps": 6870, "loss": 0.2146, "lr": 1.2720656104963569e-05, "epoch": 2.091703056768559, "percentage": 69.72, "elapsed_time": "19:02:25", "remaining_time": "8:16:05"} +{"current_steps": 4795, "total_steps": 6870, "loss": 0.2169, "lr": 1.2665372307705525e-05, "epoch": 2.093886462882096, "percentage": 69.8, "elapsed_time": "19:03:37", "remaining_time": "8:14:53"} +{"current_steps": 4800, "total_steps": 6870, "loss": 0.2203, "lr": 1.2610168120346455e-05, "epoch": 2.096069868995633, "percentage": 69.87, "elapsed_time": "19:04:48", "remaining_time": "8:13:41"} +{"current_steps": 4805, "total_steps": 6870, "loss": 0.2169, "lr": 1.2555043899184077e-05, "epoch": 2.0982532751091703, "percentage": 69.94, "elapsed_time": "19:06:00", "remaining_time": "8:12:30"} +{"current_steps": 4810, "total_steps": 6870, "loss": 0.2134, "lr": 1.2500000000000006e-05, "epoch": 2.1004366812227073, "percentage": 70.01, "elapsed_time": "19:07:11", "remaining_time": "8:11:18"} +{"current_steps": 4815, "total_steps": 6870, "loss": 0.2156, "lr": 1.2445036778057434e-05, "epoch": 2.1026200873362444, "percentage": 70.09, "elapsed_time": "19:08:23", "remaining_time": "8:10:07"} +{"current_steps": 4820, "total_steps": 6870, "loss": 0.2155, "lr": 1.2390154588098854e-05, "epoch": 2.1048034934497815, "percentage": 70.16, "elapsed_time": "19:09:34", "remaining_time": "8:08:55"} +{"current_steps": 4825, "total_steps": 6870, "loss": 0.2078, "lr": 1.233535378434376e-05, "epoch": 2.106986899563319, "percentage": 70.23, "elapsed_time": "19:10:46", "remaining_time": "8:07:44"} +{"current_steps": 4830, "total_steps": 6870, "loss": 0.2138, "lr": 1.2280634720486364e-05, "epoch": 2.109170305676856, "percentage": 70.31, "elapsed_time": "19:11:57", "remaining_time": "8:06:32"} +{"current_steps": 4835, "total_steps": 6870, "loss": 0.2098, "lr": 1.2225997749693318e-05, "epoch": 2.111353711790393, "percentage": 70.38, "elapsed_time": "19:13:09", "remaining_time": "8:05:21"} +{"current_steps": 4840, "total_steps": 6870, "loss": 0.2141, "lr": 1.2171443224601443e-05, "epoch": 2.11353711790393, "percentage": 70.45, "elapsed_time": "19:14:21", "remaining_time": "8:04:09"} +{"current_steps": 4845, "total_steps": 6870, "loss": 0.2079, "lr": 1.2116971497315402e-05, "epoch": 2.1157205240174672, "percentage": 70.52, "elapsed_time": "19:15:32", "remaining_time": "8:02:58"} +{"current_steps": 4850, "total_steps": 6870, "loss": 0.2096, "lr": 1.2062582919405502e-05, "epoch": 2.1179039301310043, "percentage": 70.6, "elapsed_time": "19:16:44", "remaining_time": "8:01:46"} +{"current_steps": 4855, "total_steps": 6870, "loss": 0.2142, "lr": 1.200827784190537e-05, "epoch": 2.1200873362445414, "percentage": 70.67, "elapsed_time": "19:17:55", "remaining_time": "8:00:34"} +{"current_steps": 4860, "total_steps": 6870, "loss": 0.2086, "lr": 1.1954056615309728e-05, "epoch": 2.1222707423580784, "percentage": 70.74, "elapsed_time": "19:19:07", "remaining_time": "7:59:23"} +{"current_steps": 4865, "total_steps": 6870, "loss": 0.2129, "lr": 1.1899919589572092e-05, "epoch": 2.1244541484716155, "percentage": 70.82, "elapsed_time": "19:20:18", "remaining_time": "7:58:11"} +{"current_steps": 4870, "total_steps": 6870, "loss": 0.2064, "lr": 1.1845867114102532e-05, "epoch": 2.126637554585153, "percentage": 70.89, "elapsed_time": "19:21:30", "remaining_time": "7:57:00"} +{"current_steps": 4875, "total_steps": 6870, "loss": 0.2226, "lr": 1.1791899537765416e-05, "epoch": 2.12882096069869, "percentage": 70.96, "elapsed_time": "19:22:42", "remaining_time": "7:55:48"} +{"current_steps": 4880, "total_steps": 6870, "loss": 0.2218, "lr": 1.173801720887717e-05, "epoch": 2.131004366812227, "percentage": 71.03, "elapsed_time": "19:23:53", "remaining_time": "7:54:37"} +{"current_steps": 4885, "total_steps": 6870, "loss": 0.1998, "lr": 1.1684220475203986e-05, "epoch": 2.1331877729257642, "percentage": 71.11, "elapsed_time": "19:25:05", "remaining_time": "7:53:25"} +{"current_steps": 4890, "total_steps": 6870, "loss": 0.2123, "lr": 1.1630509683959643e-05, "epoch": 2.1353711790393013, "percentage": 71.18, "elapsed_time": "19:26:16", "remaining_time": "7:52:14"} +{"current_steps": 4895, "total_steps": 6870, "loss": 0.2171, "lr": 1.1576885181803215e-05, "epoch": 2.1375545851528384, "percentage": 71.25, "elapsed_time": "19:27:28", "remaining_time": "7:51:02"} +{"current_steps": 4900, "total_steps": 6870, "loss": 0.2135, "lr": 1.1523347314836858e-05, "epoch": 2.1397379912663754, "percentage": 71.32, "elapsed_time": "19:28:39", "remaining_time": "7:49:51"} +{"current_steps": 4905, "total_steps": 6870, "loss": 0.2202, "lr": 1.1469896428603563e-05, "epoch": 2.1419213973799125, "percentage": 71.4, "elapsed_time": "19:29:51", "remaining_time": "7:48:39"} +{"current_steps": 4910, "total_steps": 6870, "loss": 0.2023, "lr": 1.1416532868084937e-05, "epoch": 2.14410480349345, "percentage": 71.47, "elapsed_time": "19:31:02", "remaining_time": "7:47:27"} +{"current_steps": 4915, "total_steps": 6870, "loss": 0.2155, "lr": 1.1363256977698968e-05, "epoch": 2.146288209606987, "percentage": 71.54, "elapsed_time": "19:32:14", "remaining_time": "7:46:16"} +{"current_steps": 4920, "total_steps": 6870, "loss": 0.2119, "lr": 1.1310069101297807e-05, "epoch": 2.148471615720524, "percentage": 71.62, "elapsed_time": "19:33:26", "remaining_time": "7:45:04"} +{"current_steps": 4925, "total_steps": 6870, "loss": 0.21, "lr": 1.1256969582165544e-05, "epoch": 2.150655021834061, "percentage": 71.69, "elapsed_time": "19:34:37", "remaining_time": "7:43:53"} +{"current_steps": 4930, "total_steps": 6870, "loss": 0.1967, "lr": 1.1203958763016007e-05, "epoch": 2.1528384279475983, "percentage": 71.76, "elapsed_time": "19:35:49", "remaining_time": "7:42:41"} +{"current_steps": 4935, "total_steps": 6870, "loss": 0.2095, "lr": 1.1151036985990509e-05, "epoch": 2.1550218340611353, "percentage": 71.83, "elapsed_time": "19:37:00", "remaining_time": "7:41:30"} +{"current_steps": 4940, "total_steps": 6870, "loss": 0.2144, "lr": 1.1098204592655703e-05, "epoch": 2.1572052401746724, "percentage": 71.91, "elapsed_time": "19:38:12", "remaining_time": "7:40:18"} +{"current_steps": 4945, "total_steps": 6870, "loss": 0.2094, "lr": 1.1045461924001325e-05, "epoch": 2.1593886462882095, "percentage": 71.98, "elapsed_time": "19:39:23", "remaining_time": "7:39:06"} +{"current_steps": 4950, "total_steps": 6870, "loss": 0.2073, "lr": 1.0992809320438016e-05, "epoch": 2.1615720524017465, "percentage": 72.05, "elapsed_time": "19:40:35", "remaining_time": "7:37:55"} +{"current_steps": 4955, "total_steps": 6870, "loss": 0.2095, "lr": 1.0940247121795123e-05, "epoch": 2.1637554585152836, "percentage": 72.13, "elapsed_time": "19:41:46", "remaining_time": "7:36:43"} +{"current_steps": 4960, "total_steps": 6870, "loss": 0.2172, "lr": 1.08877756673185e-05, "epoch": 2.165938864628821, "percentage": 72.2, "elapsed_time": "19:42:58", "remaining_time": "7:35:32"} +{"current_steps": 4965, "total_steps": 6870, "loss": 0.2177, "lr": 1.0835395295668327e-05, "epoch": 2.168122270742358, "percentage": 72.27, "elapsed_time": "19:44:09", "remaining_time": "7:34:20"} +{"current_steps": 4970, "total_steps": 6870, "loss": 0.2051, "lr": 1.0783106344916913e-05, "epoch": 2.1703056768558953, "percentage": 72.34, "elapsed_time": "19:45:21", "remaining_time": "7:33:09"} +{"current_steps": 4975, "total_steps": 6870, "loss": 0.2222, "lr": 1.073090915254652e-05, "epoch": 2.1724890829694323, "percentage": 72.42, "elapsed_time": "19:46:32", "remaining_time": "7:31:57"} +{"current_steps": 4980, "total_steps": 6870, "loss": 0.2005, "lr": 1.06788040554472e-05, "epoch": 2.1746724890829694, "percentage": 72.49, "elapsed_time": "19:47:44", "remaining_time": "7:30:46"} +{"current_steps": 4985, "total_steps": 6870, "loss": 0.203, "lr": 1.0626791389914573e-05, "epoch": 2.1768558951965065, "percentage": 72.56, "elapsed_time": "19:48:56", "remaining_time": "7:29:34"} +{"current_steps": 4990, "total_steps": 6870, "loss": 0.214, "lr": 1.0574871491647723e-05, "epoch": 2.1790393013100435, "percentage": 72.63, "elapsed_time": "19:50:07", "remaining_time": "7:28:23"} +{"current_steps": 4995, "total_steps": 6870, "loss": 0.2125, "lr": 1.0523044695746975e-05, "epoch": 2.1812227074235806, "percentage": 72.71, "elapsed_time": "19:51:19", "remaining_time": "7:27:11"} +{"current_steps": 5000, "total_steps": 6870, "loss": 0.2119, "lr": 1.0471311336711772e-05, "epoch": 2.183406113537118, "percentage": 72.78, "elapsed_time": "19:52:30", "remaining_time": "7:26:00"} +{"current_steps": 5005, "total_steps": 6870, "loss": 0.2168, "lr": 1.0419671748438487e-05, "epoch": 2.185589519650655, "percentage": 72.85, "elapsed_time": "19:53:42", "remaining_time": "7:24:48"} +{"current_steps": 5010, "total_steps": 6870, "loss": 0.211, "lr": 1.0368126264218282e-05, "epoch": 2.1877729257641922, "percentage": 72.93, "elapsed_time": "19:54:54", "remaining_time": "7:23:36"} +{"current_steps": 5015, "total_steps": 6870, "loss": 0.2057, "lr": 1.0316675216734952e-05, "epoch": 2.1899563318777293, "percentage": 73.0, "elapsed_time": "19:56:05", "remaining_time": "7:22:25"} +{"current_steps": 5020, "total_steps": 6870, "loss": 0.2229, "lr": 1.026531893806279e-05, "epoch": 2.1921397379912664, "percentage": 73.07, "elapsed_time": "19:57:17", "remaining_time": "7:21:13"} +{"current_steps": 5025, "total_steps": 6870, "loss": 0.2159, "lr": 1.0214057759664424e-05, "epoch": 2.1943231441048034, "percentage": 73.14, "elapsed_time": "19:58:28", "remaining_time": "7:20:02"} +{"current_steps": 5030, "total_steps": 6870, "loss": 0.223, "lr": 1.0162892012388703e-05, "epoch": 2.1965065502183405, "percentage": 73.22, "elapsed_time": "19:59:40", "remaining_time": "7:18:50"} +{"current_steps": 5035, "total_steps": 6870, "loss": 0.2127, "lr": 1.0111822026468515e-05, "epoch": 2.1986899563318776, "percentage": 73.29, "elapsed_time": "20:00:51", "remaining_time": "7:17:39"} +{"current_steps": 5040, "total_steps": 6870, "loss": 0.2133, "lr": 1.0060848131518716e-05, "epoch": 2.2008733624454146, "percentage": 73.36, "elapsed_time": "20:02:03", "remaining_time": "7:16:27"} +{"current_steps": 5045, "total_steps": 6870, "loss": 0.2137, "lr": 1.0009970656533965e-05, "epoch": 2.203056768558952, "percentage": 73.44, "elapsed_time": "20:03:15", "remaining_time": "7:15:16"} +{"current_steps": 5050, "total_steps": 6870, "loss": 0.2161, "lr": 9.95918992988661e-06, "epoch": 2.2052401746724892, "percentage": 73.51, "elapsed_time": "20:04:26", "remaining_time": "7:14:04"} +{"current_steps": 5055, "total_steps": 6870, "loss": 0.2097, "lr": 9.908506279324564e-06, "epoch": 2.2074235807860263, "percentage": 73.58, "elapsed_time": "20:05:38", "remaining_time": "7:12:53"} +{"current_steps": 5060, "total_steps": 6870, "loss": 0.2065, "lr": 9.857920031969198e-06, "epoch": 2.2096069868995634, "percentage": 73.65, "elapsed_time": "20:06:49", "remaining_time": "7:11:41"} +{"current_steps": 5065, "total_steps": 6870, "loss": 0.2313, "lr": 9.807431514313225e-06, "epoch": 2.2117903930131004, "percentage": 73.73, "elapsed_time": "20:08:01", "remaining_time": "7:10:29"} +{"current_steps": 5070, "total_steps": 6870, "loss": 0.2192, "lr": 9.757041052218587e-06, "epoch": 2.2139737991266375, "percentage": 73.8, "elapsed_time": "20:09:12", "remaining_time": "7:09:18"} +{"current_steps": 5075, "total_steps": 6870, "loss": 0.2092, "lr": 9.70674897091437e-06, "epoch": 2.2161572052401746, "percentage": 73.87, "elapsed_time": "20:10:24", "remaining_time": "7:08:06"} +{"current_steps": 5080, "total_steps": 6870, "loss": 0.2027, "lr": 9.656555594994687e-06, "epoch": 2.2183406113537116, "percentage": 73.94, "elapsed_time": "20:11:36", "remaining_time": "7:06:55"} +{"current_steps": 5085, "total_steps": 6870, "loss": 0.2016, "lr": 9.606461248416573e-06, "epoch": 2.2205240174672487, "percentage": 74.02, "elapsed_time": "20:12:47", "remaining_time": "7:05:43"} +{"current_steps": 5090, "total_steps": 6870, "loss": 0.2098, "lr": 9.556466254497931e-06, "epoch": 2.222707423580786, "percentage": 74.09, "elapsed_time": "20:13:59", "remaining_time": "7:04:32"} +{"current_steps": 5095, "total_steps": 6870, "loss": 0.2081, "lr": 9.506570935915419e-06, "epoch": 2.2248908296943233, "percentage": 74.16, "elapsed_time": "20:15:10", "remaining_time": "7:03:20"} +{"current_steps": 5100, "total_steps": 6870, "loss": 0.2082, "lr": 9.456775614702364e-06, "epoch": 2.2270742358078603, "percentage": 74.24, "elapsed_time": "20:16:22", "remaining_time": "7:02:09"} +{"current_steps": 5105, "total_steps": 6870, "loss": 0.2106, "lr": 9.407080612246697e-06, "epoch": 2.2292576419213974, "percentage": 74.31, "elapsed_time": "20:17:33", "remaining_time": "7:00:57"} +{"current_steps": 5110, "total_steps": 6870, "loss": 0.2252, "lr": 9.357486249288891e-06, "epoch": 2.2314410480349345, "percentage": 74.38, "elapsed_time": "20:18:45", "remaining_time": "6:59:46"} +{"current_steps": 5115, "total_steps": 6870, "loss": 0.2022, "lr": 9.307992845919844e-06, "epoch": 2.2336244541484715, "percentage": 74.45, "elapsed_time": "20:19:57", "remaining_time": "6:58:34"} +{"current_steps": 5120, "total_steps": 6870, "loss": 0.2075, "lr": 9.258600721578872e-06, "epoch": 2.2358078602620086, "percentage": 74.53, "elapsed_time": "20:21:08", "remaining_time": "6:57:22"} +{"current_steps": 5125, "total_steps": 6870, "loss": 0.2187, "lr": 9.209310195051582e-06, "epoch": 2.2379912663755457, "percentage": 74.6, "elapsed_time": "20:22:20", "remaining_time": "6:56:11"} +{"current_steps": 5130, "total_steps": 6870, "loss": 0.1939, "lr": 9.160121584467878e-06, "epoch": 2.2401746724890828, "percentage": 74.67, "elapsed_time": "20:23:31", "remaining_time": "6:54:59"} +{"current_steps": 5135, "total_steps": 6870, "loss": 0.2153, "lr": 9.111035207299878e-06, "epoch": 2.2423580786026203, "percentage": 74.75, "elapsed_time": "20:24:43", "remaining_time": "6:53:48"} +{"current_steps": 5140, "total_steps": 6870, "loss": 0.2155, "lr": 9.062051380359857e-06, "epoch": 2.2445414847161573, "percentage": 74.82, "elapsed_time": "20:25:54", "remaining_time": "6:52:36"} +{"current_steps": 5145, "total_steps": 6870, "loss": 0.219, "lr": 9.01317041979822e-06, "epoch": 2.2467248908296944, "percentage": 74.89, "elapsed_time": "20:27:06", "remaining_time": "6:51:25"} +{"current_steps": 5150, "total_steps": 6870, "loss": 0.1981, "lr": 8.964392641101455e-06, "epoch": 2.2489082969432315, "percentage": 74.96, "elapsed_time": "20:28:17", "remaining_time": "6:50:13"} +{"current_steps": 5155, "total_steps": 6870, "loss": 0.2047, "lr": 8.915718359090086e-06, "epoch": 2.2510917030567685, "percentage": 75.04, "elapsed_time": "20:29:29", "remaining_time": "6:49:02"} +{"current_steps": 5160, "total_steps": 6870, "loss": 0.2228, "lr": 8.867147887916663e-06, "epoch": 2.2532751091703056, "percentage": 75.11, "elapsed_time": "20:30:40", "remaining_time": "6:47:50"} +{"current_steps": 5165, "total_steps": 6870, "loss": 0.2115, "lr": 8.818681541063714e-06, "epoch": 2.2554585152838427, "percentage": 75.18, "elapsed_time": "20:31:52", "remaining_time": "6:46:38"} +{"current_steps": 5170, "total_steps": 6870, "loss": 0.2013, "lr": 8.770319631341745e-06, "epoch": 2.2576419213973797, "percentage": 75.25, "elapsed_time": "20:33:04", "remaining_time": "6:45:27"} +{"current_steps": 5175, "total_steps": 6870, "loss": 0.1977, "lr": 8.722062470887169e-06, "epoch": 2.2598253275109172, "percentage": 75.33, "elapsed_time": "20:34:15", "remaining_time": "6:44:15"} +{"current_steps": 5180, "total_steps": 6870, "loss": 0.213, "lr": 8.673910371160368e-06, "epoch": 2.2620087336244543, "percentage": 75.4, "elapsed_time": "20:35:27", "remaining_time": "6:43:04"} +{"current_steps": 5185, "total_steps": 6870, "loss": 0.2104, "lr": 8.625863642943623e-06, "epoch": 2.2641921397379914, "percentage": 75.47, "elapsed_time": "20:36:38", "remaining_time": "6:41:52"} +{"current_steps": 5190, "total_steps": 6870, "loss": 0.2088, "lr": 8.577922596339142e-06, "epoch": 2.2663755458515285, "percentage": 75.55, "elapsed_time": "20:37:50", "remaining_time": "6:40:41"} +{"current_steps": 5195, "total_steps": 6870, "loss": 0.212, "lr": 8.530087540767029e-06, "epoch": 2.2685589519650655, "percentage": 75.62, "elapsed_time": "20:39:01", "remaining_time": "6:39:29"} +{"current_steps": 5200, "total_steps": 6870, "loss": 0.209, "lr": 8.482358784963312e-06, "epoch": 2.2707423580786026, "percentage": 75.69, "elapsed_time": "20:40:13", "remaining_time": "6:38:18"} +{"current_steps": 5205, "total_steps": 6870, "loss": 0.1947, "lr": 8.43473663697794e-06, "epoch": 2.2729257641921397, "percentage": 75.76, "elapsed_time": "20:41:25", "remaining_time": "6:37:06"} +{"current_steps": 5210, "total_steps": 6870, "loss": 0.2036, "lr": 8.387221404172796e-06, "epoch": 2.2751091703056767, "percentage": 75.84, "elapsed_time": "20:42:36", "remaining_time": "6:35:55"} +{"current_steps": 5215, "total_steps": 6870, "loss": 0.2144, "lr": 8.339813393219714e-06, "epoch": 2.277292576419214, "percentage": 75.91, "elapsed_time": "20:43:48", "remaining_time": "6:34:43"} +{"current_steps": 5220, "total_steps": 6870, "loss": 0.2204, "lr": 8.292512910098508e-06, "epoch": 2.279475982532751, "percentage": 75.98, "elapsed_time": "20:44:59", "remaining_time": "6:33:32"} +{"current_steps": 5225, "total_steps": 6870, "loss": 0.2199, "lr": 8.245320260094955e-06, "epoch": 2.2816593886462884, "percentage": 76.06, "elapsed_time": "20:46:11", "remaining_time": "6:32:20"} +{"current_steps": 5230, "total_steps": 6870, "loss": 0.2169, "lr": 8.198235747798893e-06, "epoch": 2.2838427947598254, "percentage": 76.13, "elapsed_time": "20:47:23", "remaining_time": "6:31:08"} +{"current_steps": 5235, "total_steps": 6870, "loss": 0.2158, "lr": 8.151259677102211e-06, "epoch": 2.2860262008733625, "percentage": 76.2, "elapsed_time": "20:48:34", "remaining_time": "6:29:57"} +{"current_steps": 5240, "total_steps": 6870, "loss": 0.2525, "lr": 8.104392351196885e-06, "epoch": 2.2882096069868996, "percentage": 76.27, "elapsed_time": "20:49:46", "remaining_time": "6:28:45"} +{"current_steps": 5245, "total_steps": 6870, "loss": 0.2176, "lr": 8.057634072573048e-06, "epoch": 2.2903930131004366, "percentage": 76.35, "elapsed_time": "20:50:57", "remaining_time": "6:27:34"} +{"current_steps": 5250, "total_steps": 6870, "loss": 0.2159, "lr": 8.010985143017008e-06, "epoch": 2.2925764192139737, "percentage": 76.42, "elapsed_time": "20:52:09", "remaining_time": "6:26:22"} +{"current_steps": 5255, "total_steps": 6870, "loss": 0.2152, "lr": 7.964445863609329e-06, "epoch": 2.2947598253275108, "percentage": 76.49, "elapsed_time": "20:53:20", "remaining_time": "6:25:11"} +{"current_steps": 5260, "total_steps": 6870, "loss": 0.1989, "lr": 7.91801653472286e-06, "epoch": 2.2969432314410483, "percentage": 76.56, "elapsed_time": "20:54:32", "remaining_time": "6:23:59"} +{"current_steps": 5265, "total_steps": 6870, "loss": 0.2282, "lr": 7.871697456020819e-06, "epoch": 2.2991266375545854, "percentage": 76.64, "elapsed_time": "20:55:43", "remaining_time": "6:22:48"} +{"current_steps": 5270, "total_steps": 6870, "loss": 0.2137, "lr": 7.825488926454851e-06, "epoch": 2.3013100436681224, "percentage": 76.71, "elapsed_time": "20:56:55", "remaining_time": "6:21:36"} +{"current_steps": 5275, "total_steps": 6870, "loss": 0.2092, "lr": 7.77939124426308e-06, "epoch": 2.3034934497816595, "percentage": 76.78, "elapsed_time": "20:58:06", "remaining_time": "6:20:24"} +{"current_steps": 5280, "total_steps": 6870, "loss": 0.2093, "lr": 7.733404706968215e-06, "epoch": 2.3056768558951966, "percentage": 76.86, "elapsed_time": "20:59:18", "remaining_time": "6:19:13"} +{"current_steps": 5285, "total_steps": 6870, "loss": 0.216, "lr": 7.687529611375616e-06, "epoch": 2.3078602620087336, "percentage": 76.93, "elapsed_time": "21:00:29", "remaining_time": "6:18:01"} +{"current_steps": 5290, "total_steps": 6870, "loss": 0.2039, "lr": 7.64176625357138e-06, "epoch": 2.3100436681222707, "percentage": 77.0, "elapsed_time": "21:01:41", "remaining_time": "6:16:50"} +{"current_steps": 5295, "total_steps": 6870, "loss": 0.2119, "lr": 7.596114928920422e-06, "epoch": 2.3122270742358078, "percentage": 77.07, "elapsed_time": "21:02:52", "remaining_time": "6:15:38"} +{"current_steps": 5300, "total_steps": 6870, "loss": 0.2013, "lr": 7.55057593206458e-06, "epoch": 2.314410480349345, "percentage": 77.15, "elapsed_time": "21:04:04", "remaining_time": "6:14:27"} +{"current_steps": 5305, "total_steps": 6870, "loss": 0.2138, "lr": 7.505149556920699e-06, "epoch": 2.316593886462882, "percentage": 77.22, "elapsed_time": "21:05:16", "remaining_time": "6:13:15"} +{"current_steps": 5310, "total_steps": 6870, "loss": 0.2151, "lr": 7.459836096678757e-06, "epoch": 2.3187772925764194, "percentage": 77.29, "elapsed_time": "21:06:27", "remaining_time": "6:12:04"} +{"current_steps": 5315, "total_steps": 6870, "loss": 0.2145, "lr": 7.41463584379995e-06, "epoch": 2.3209606986899565, "percentage": 77.37, "elapsed_time": "21:07:39", "remaining_time": "6:10:52"} +{"current_steps": 5320, "total_steps": 6870, "loss": 0.2118, "lr": 7.369549090014821e-06, "epoch": 2.3231441048034935, "percentage": 77.44, "elapsed_time": "21:08:50", "remaining_time": "6:09:40"} +{"current_steps": 5325, "total_steps": 6870, "loss": 0.216, "lr": 7.32457612632135e-06, "epoch": 2.3253275109170306, "percentage": 77.51, "elapsed_time": "21:10:02", "remaining_time": "6:08:29"} +{"current_steps": 5330, "total_steps": 6870, "loss": 0.2167, "lr": 7.279717242983117e-06, "epoch": 2.3275109170305677, "percentage": 77.58, "elapsed_time": "21:11:13", "remaining_time": "6:07:17"} +{"current_steps": 5335, "total_steps": 6870, "loss": 0.2073, "lr": 7.234972729527401e-06, "epoch": 2.3296943231441047, "percentage": 77.66, "elapsed_time": "21:12:25", "remaining_time": "6:06:06"} +{"current_steps": 5340, "total_steps": 6870, "loss": 0.2012, "lr": 7.190342874743311e-06, "epoch": 2.331877729257642, "percentage": 77.73, "elapsed_time": "21:13:36", "remaining_time": "6:04:54"} +{"current_steps": 5345, "total_steps": 6870, "loss": 0.1967, "lr": 7.145827966679936e-06, "epoch": 2.334061135371179, "percentage": 77.8, "elapsed_time": "21:14:48", "remaining_time": "6:03:43"} +{"current_steps": 5350, "total_steps": 6870, "loss": 0.2176, "lr": 7.101428292644477e-06, "epoch": 2.3362445414847164, "percentage": 77.87, "elapsed_time": "21:15:59", "remaining_time": "6:02:31"} +{"current_steps": 5355, "total_steps": 6870, "loss": 0.2016, "lr": 7.057144139200378e-06, "epoch": 2.3384279475982535, "percentage": 77.95, "elapsed_time": "21:17:11", "remaining_time": "6:01:20"} +{"current_steps": 5360, "total_steps": 6870, "loss": 0.1983, "lr": 7.0129757921655355e-06, "epoch": 2.3406113537117905, "percentage": 78.02, "elapsed_time": "21:18:23", "remaining_time": "6:00:08"} +{"current_steps": 5365, "total_steps": 6870, "loss": 0.208, "lr": 6.968923536610356e-06, "epoch": 2.3427947598253276, "percentage": 78.09, "elapsed_time": "21:19:34", "remaining_time": "5:58:56"} +{"current_steps": 5370, "total_steps": 6870, "loss": 0.194, "lr": 6.924987656856002e-06, "epoch": 2.3449781659388647, "percentage": 78.17, "elapsed_time": "21:20:46", "remaining_time": "5:57:45"} +{"current_steps": 5375, "total_steps": 6870, "loss": 0.2111, "lr": 6.881168436472524e-06, "epoch": 2.3471615720524017, "percentage": 78.24, "elapsed_time": "21:21:57", "remaining_time": "5:56:33"} +{"current_steps": 5380, "total_steps": 6870, "loss": 0.1942, "lr": 6.837466158277025e-06, "epoch": 2.349344978165939, "percentage": 78.31, "elapsed_time": "21:23:09", "remaining_time": "5:55:22"} +{"current_steps": 5385, "total_steps": 6870, "loss": 0.2149, "lr": 6.7938811043318475e-06, "epoch": 2.351528384279476, "percentage": 78.38, "elapsed_time": "21:24:20", "remaining_time": "5:54:10"} +{"current_steps": 5390, "total_steps": 6870, "loss": 0.2066, "lr": 6.750413555942742e-06, "epoch": 2.353711790393013, "percentage": 78.46, "elapsed_time": "21:25:32", "remaining_time": "5:52:59"} +{"current_steps": 5395, "total_steps": 6870, "loss": 0.217, "lr": 6.707063793657065e-06, "epoch": 2.35589519650655, "percentage": 78.53, "elapsed_time": "21:26:44", "remaining_time": "5:51:47"} +{"current_steps": 5400, "total_steps": 6870, "loss": 0.2198, "lr": 6.663832097261952e-06, "epoch": 2.3580786026200875, "percentage": 78.6, "elapsed_time": "21:27:55", "remaining_time": "5:50:36"} +{"current_steps": 5405, "total_steps": 6870, "loss": 0.2032, "lr": 6.620718745782525e-06, "epoch": 2.3602620087336246, "percentage": 78.68, "elapsed_time": "21:29:07", "remaining_time": "5:49:24"} +{"current_steps": 5410, "total_steps": 6870, "loss": 0.2146, "lr": 6.577724017480094e-06, "epoch": 2.3624454148471616, "percentage": 78.75, "elapsed_time": "21:30:18", "remaining_time": "5:48:13"} +{"current_steps": 5415, "total_steps": 6870, "loss": 0.199, "lr": 6.534848189850329e-06, "epoch": 2.3646288209606987, "percentage": 78.82, "elapsed_time": "21:31:30", "remaining_time": "5:47:01"} +{"current_steps": 5420, "total_steps": 6870, "loss": 0.2034, "lr": 6.492091539621517e-06, "epoch": 2.3668122270742358, "percentage": 78.89, "elapsed_time": "21:32:41", "remaining_time": "5:45:49"} +{"current_steps": 5425, "total_steps": 6870, "loss": 0.2018, "lr": 6.449454342752748e-06, "epoch": 2.368995633187773, "percentage": 78.97, "elapsed_time": "21:33:53", "remaining_time": "5:44:38"} +{"current_steps": 5430, "total_steps": 6870, "loss": 0.2038, "lr": 6.406936874432134e-06, "epoch": 2.37117903930131, "percentage": 79.04, "elapsed_time": "21:35:04", "remaining_time": "5:43:26"} +{"current_steps": 5435, "total_steps": 6870, "loss": 0.2018, "lr": 6.364539409075043e-06, "epoch": 2.373362445414847, "percentage": 79.11, "elapsed_time": "21:36:16", "remaining_time": "5:42:15"} +{"current_steps": 5440, "total_steps": 6870, "loss": 0.2092, "lr": 6.322262220322314e-06, "epoch": 2.3755458515283845, "percentage": 79.18, "elapsed_time": "21:37:28", "remaining_time": "5:41:03"} +{"current_steps": 5445, "total_steps": 6870, "loss": 0.2036, "lr": 6.280105581038509e-06, "epoch": 2.3777292576419216, "percentage": 79.26, "elapsed_time": "21:38:39", "remaining_time": "5:39:52"} +{"current_steps": 5450, "total_steps": 6870, "loss": 0.208, "lr": 6.238069763310136e-06, "epoch": 2.3799126637554586, "percentage": 79.33, "elapsed_time": "21:39:51", "remaining_time": "5:38:40"} +{"current_steps": 5455, "total_steps": 6870, "loss": 0.2072, "lr": 6.1961550384438984e-06, "epoch": 2.3820960698689957, "percentage": 79.4, "elapsed_time": "21:41:03", "remaining_time": "5:37:29"} +{"current_steps": 5460, "total_steps": 6870, "loss": 0.2166, "lr": 6.154361676964956e-06, "epoch": 2.3842794759825328, "percentage": 79.48, "elapsed_time": "21:42:14", "remaining_time": "5:36:17"} +{"current_steps": 5465, "total_steps": 6870, "loss": 0.2021, "lr": 6.112689948615144e-06, "epoch": 2.38646288209607, "percentage": 79.55, "elapsed_time": "21:43:26", "remaining_time": "5:35:06"} +{"current_steps": 5470, "total_steps": 6870, "loss": 0.2018, "lr": 6.071140122351276e-06, "epoch": 2.388646288209607, "percentage": 79.62, "elapsed_time": "21:44:37", "remaining_time": "5:33:54"} +{"current_steps": 5475, "total_steps": 6870, "loss": 0.2057, "lr": 6.029712466343382e-06, "epoch": 2.390829694323144, "percentage": 79.69, "elapsed_time": "21:45:49", "remaining_time": "5:32:42"} +{"current_steps": 5480, "total_steps": 6870, "loss": 0.2102, "lr": 5.988407247972982e-06, "epoch": 2.393013100436681, "percentage": 79.77, "elapsed_time": "21:47:00", "remaining_time": "5:31:31"} +{"current_steps": 5485, "total_steps": 6870, "loss": 0.2143, "lr": 5.947224733831364e-06, "epoch": 2.395196506550218, "percentage": 79.84, "elapsed_time": "21:48:12", "remaining_time": "5:30:19"} +{"current_steps": 5490, "total_steps": 6870, "loss": 0.207, "lr": 5.906165189717855e-06, "epoch": 2.3973799126637556, "percentage": 79.91, "elapsed_time": "21:49:23", "remaining_time": "5:29:08"} +{"current_steps": 5495, "total_steps": 6870, "loss": 0.2095, "lr": 5.8652288806381215e-06, "epoch": 2.3995633187772927, "percentage": 79.99, "elapsed_time": "21:50:35", "remaining_time": "5:27:56"} +{"current_steps": 5500, "total_steps": 6870, "loss": 0.1987, "lr": 5.824416070802438e-06, "epoch": 2.4017467248908297, "percentage": 80.06, "elapsed_time": "21:51:47", "remaining_time": "5:26:45"} +{"current_steps": 5505, "total_steps": 6870, "loss": 0.1982, "lr": 5.7837270236239995e-06, "epoch": 2.403930131004367, "percentage": 80.13, "elapsed_time": "21:52:58", "remaining_time": "5:25:33"} +{"current_steps": 5510, "total_steps": 6870, "loss": 0.2043, "lr": 5.74316200171722e-06, "epoch": 2.406113537117904, "percentage": 80.2, "elapsed_time": "21:54:10", "remaining_time": "5:24:22"} +{"current_steps": 5515, "total_steps": 6870, "loss": 0.2032, "lr": 5.70272126689601e-06, "epoch": 2.408296943231441, "percentage": 80.28, "elapsed_time": "21:55:21", "remaining_time": "5:23:10"} +{"current_steps": 5520, "total_steps": 6870, "loss": 0.2077, "lr": 5.662405080172134e-06, "epoch": 2.410480349344978, "percentage": 80.35, "elapsed_time": "21:56:33", "remaining_time": "5:21:58"} +{"current_steps": 5525, "total_steps": 6870, "loss": 0.2235, "lr": 5.622213701753487e-06, "epoch": 2.4126637554585155, "percentage": 80.42, "elapsed_time": "21:57:44", "remaining_time": "5:20:47"} +{"current_steps": 5530, "total_steps": 6870, "loss": 0.2092, "lr": 5.582147391042433e-06, "epoch": 2.4148471615720526, "percentage": 80.49, "elapsed_time": "21:58:56", "remaining_time": "5:19:35"} +{"current_steps": 5535, "total_steps": 6870, "loss": 0.2079, "lr": 5.542206406634129e-06, "epoch": 2.4170305676855897, "percentage": 80.57, "elapsed_time": "22:00:07", "remaining_time": "5:18:24"} +{"current_steps": 5540, "total_steps": 6870, "loss": 0.2111, "lr": 5.5023910063148445e-06, "epoch": 2.4192139737991267, "percentage": 80.64, "elapsed_time": "22:01:19", "remaining_time": "5:17:12"} +{"current_steps": 5545, "total_steps": 6870, "loss": 0.1979, "lr": 5.46270144706032e-06, "epoch": 2.421397379912664, "percentage": 80.71, "elapsed_time": "22:02:30", "remaining_time": "5:16:01"} +{"current_steps": 5550, "total_steps": 6870, "loss": 0.2136, "lr": 5.423137985034085e-06, "epoch": 2.423580786026201, "percentage": 80.79, "elapsed_time": "22:03:42", "remaining_time": "5:14:49"} +{"current_steps": 5555, "total_steps": 6870, "loss": 0.2006, "lr": 5.3837008755858145e-06, "epoch": 2.425764192139738, "percentage": 80.86, "elapsed_time": "22:04:54", "remaining_time": "5:13:38"} +{"current_steps": 5560, "total_steps": 6870, "loss": 0.2068, "lr": 5.3443903732496966e-06, "epoch": 2.427947598253275, "percentage": 80.93, "elapsed_time": "22:06:05", "remaining_time": "5:12:26"} +{"current_steps": 5565, "total_steps": 6870, "loss": 0.209, "lr": 5.305206731742748e-06, "epoch": 2.430131004366812, "percentage": 81.0, "elapsed_time": "22:07:17", "remaining_time": "5:11:15"} +{"current_steps": 5570, "total_steps": 6870, "loss": 0.2158, "lr": 5.266150203963224e-06, "epoch": 2.432314410480349, "percentage": 81.08, "elapsed_time": "22:08:28", "remaining_time": "5:10:03"} +{"current_steps": 5575, "total_steps": 6870, "loss": 0.2052, "lr": 5.227221041988956e-06, "epoch": 2.4344978165938866, "percentage": 81.15, "elapsed_time": "22:09:40", "remaining_time": "5:08:51"} +{"current_steps": 5580, "total_steps": 6870, "loss": 0.2046, "lr": 5.188419497075739e-06, "epoch": 2.4366812227074237, "percentage": 81.22, "elapsed_time": "22:10:52", "remaining_time": "5:07:40"} +{"current_steps": 5585, "total_steps": 6870, "loss": 0.2197, "lr": 5.1497458196556995e-06, "epoch": 2.438864628820961, "percentage": 81.3, "elapsed_time": "22:12:03", "remaining_time": "5:06:28"} +{"current_steps": 5590, "total_steps": 6870, "loss": 0.2089, "lr": 5.111200259335688e-06, "epoch": 2.441048034934498, "percentage": 81.37, "elapsed_time": "22:13:15", "remaining_time": "5:05:17"} +{"current_steps": 5595, "total_steps": 6870, "loss": 0.2243, "lr": 5.072783064895664e-06, "epoch": 2.443231441048035, "percentage": 81.44, "elapsed_time": "22:14:26", "remaining_time": "5:04:05"} +{"current_steps": 5600, "total_steps": 6870, "loss": 0.2195, "lr": 5.034494484287086e-06, "epoch": 2.445414847161572, "percentage": 81.51, "elapsed_time": "22:15:38", "remaining_time": "5:02:54"} +{"current_steps": 5605, "total_steps": 6870, "loss": 0.2036, "lr": 4.996334764631322e-06, "epoch": 2.447598253275109, "percentage": 81.59, "elapsed_time": "22:16:50", "remaining_time": "5:01:42"} +{"current_steps": 5610, "total_steps": 6870, "loss": 0.2077, "lr": 4.958304152218046e-06, "epoch": 2.449781659388646, "percentage": 81.66, "elapsed_time": "22:18:01", "remaining_time": "5:00:31"} +{"current_steps": 5615, "total_steps": 6870, "loss": 0.2142, "lr": 4.920402892503648e-06, "epoch": 2.4519650655021836, "percentage": 81.73, "elapsed_time": "22:19:13", "remaining_time": "4:59:19"} +{"current_steps": 5620, "total_steps": 6870, "loss": 0.2056, "lr": 4.882631230109655e-06, "epoch": 2.4541484716157207, "percentage": 81.8, "elapsed_time": "22:20:24", "remaining_time": "4:58:08"} +{"current_steps": 5625, "total_steps": 6870, "loss": 0.2044, "lr": 4.844989408821149e-06, "epoch": 2.4563318777292578, "percentage": 81.88, "elapsed_time": "22:21:36", "remaining_time": "4:56:56"} +{"current_steps": 5630, "total_steps": 6870, "loss": 0.1996, "lr": 4.807477671585195e-06, "epoch": 2.458515283842795, "percentage": 81.95, "elapsed_time": "22:22:47", "remaining_time": "4:55:44"} +{"current_steps": 5635, "total_steps": 6870, "loss": 0.203, "lr": 4.770096260509274e-06, "epoch": 2.460698689956332, "percentage": 82.02, "elapsed_time": "22:23:59", "remaining_time": "4:54:33"} +{"current_steps": 5640, "total_steps": 6870, "loss": 0.2135, "lr": 4.732845416859713e-06, "epoch": 2.462882096069869, "percentage": 82.1, "elapsed_time": "22:25:11", "remaining_time": "4:53:21"} +{"current_steps": 5645, "total_steps": 6870, "loss": 0.2066, "lr": 4.695725381060137e-06, "epoch": 2.465065502183406, "percentage": 82.17, "elapsed_time": "22:26:22", "remaining_time": "4:52:10"} +{"current_steps": 5650, "total_steps": 6870, "loss": 0.2214, "lr": 4.658736392689922e-06, "epoch": 2.467248908296943, "percentage": 82.24, "elapsed_time": "22:27:34", "remaining_time": "4:50:58"} +{"current_steps": 5655, "total_steps": 6870, "loss": 0.213, "lr": 4.6218786904826175e-06, "epoch": 2.46943231441048, "percentage": 82.31, "elapsed_time": "22:28:45", "remaining_time": "4:49:47"} +{"current_steps": 5660, "total_steps": 6870, "loss": 0.2195, "lr": 4.58515251232445e-06, "epoch": 2.4716157205240172, "percentage": 82.39, "elapsed_time": "22:29:57", "remaining_time": "4:48:35"} +{"current_steps": 5665, "total_steps": 6870, "loss": 0.2061, "lr": 4.548558095252758e-06, "epoch": 2.4737991266375547, "percentage": 82.46, "elapsed_time": "22:31:08", "remaining_time": "4:47:24"} +{"current_steps": 5670, "total_steps": 6870, "loss": 0.2116, "lr": 4.512095675454472e-06, "epoch": 2.475982532751092, "percentage": 82.53, "elapsed_time": "22:32:20", "remaining_time": "4:46:12"} +{"current_steps": 5675, "total_steps": 6870, "loss": 0.2025, "lr": 4.475765488264588e-06, "epoch": 2.478165938864629, "percentage": 82.61, "elapsed_time": "22:33:32", "remaining_time": "4:45:01"} +{"current_steps": 5680, "total_steps": 6870, "loss": 0.2016, "lr": 4.439567768164651e-06, "epoch": 2.480349344978166, "percentage": 82.68, "elapsed_time": "22:34:43", "remaining_time": "4:43:49"} +{"current_steps": 5685, "total_steps": 6870, "loss": 0.2121, "lr": 4.4035027487812325e-06, "epoch": 2.482532751091703, "percentage": 82.75, "elapsed_time": "22:35:55", "remaining_time": "4:42:37"} +{"current_steps": 5690, "total_steps": 6870, "loss": 0.1996, "lr": 4.367570662884438e-06, "epoch": 2.48471615720524, "percentage": 82.82, "elapsed_time": "22:37:06", "remaining_time": "4:41:26"} +{"current_steps": 5695, "total_steps": 6870, "loss": 0.2096, "lr": 4.331771742386395e-06, "epoch": 2.486899563318777, "percentage": 82.9, "elapsed_time": "22:38:18", "remaining_time": "4:40:14"} +{"current_steps": 5700, "total_steps": 6870, "loss": 0.2061, "lr": 4.296106218339757e-06, "epoch": 2.489082969432314, "percentage": 82.97, "elapsed_time": "22:39:29", "remaining_time": "4:39:03"} +{"current_steps": 5705, "total_steps": 6870, "loss": 0.2023, "lr": 4.260574320936203e-06, "epoch": 2.4912663755458517, "percentage": 83.04, "elapsed_time": "22:40:41", "remaining_time": "4:37:51"} +{"current_steps": 5710, "total_steps": 6870, "loss": 0.2137, "lr": 4.225176279504975e-06, "epoch": 2.493449781659389, "percentage": 83.11, "elapsed_time": "22:41:52", "remaining_time": "4:36:40"} +{"current_steps": 5715, "total_steps": 6870, "loss": 0.2037, "lr": 4.1899123225113825e-06, "epoch": 2.495633187772926, "percentage": 83.19, "elapsed_time": "22:43:04", "remaining_time": "4:35:28"} +{"current_steps": 5720, "total_steps": 6870, "loss": 0.2054, "lr": 4.154782677555324e-06, "epoch": 2.497816593886463, "percentage": 83.26, "elapsed_time": "22:44:16", "remaining_time": "4:34:17"} +{"current_steps": 5725, "total_steps": 6870, "loss": 0.2055, "lr": 4.119787571369829e-06, "epoch": 2.5, "percentage": 83.33, "elapsed_time": "22:45:27", "remaining_time": "4:33:05"} +{"current_steps": 5730, "total_steps": 6870, "loss": 0.1971, "lr": 4.084927229819591e-06, "epoch": 2.502183406113537, "percentage": 83.41, "elapsed_time": "22:46:39", "remaining_time": "4:31:54"} +{"current_steps": 5735, "total_steps": 6870, "loss": 0.1973, "lr": 4.050201877899501e-06, "epoch": 2.504366812227074, "percentage": 83.48, "elapsed_time": "22:47:50", "remaining_time": "4:30:42"} +{"current_steps": 5740, "total_steps": 6870, "loss": 0.2017, "lr": 4.0156117397332074e-06, "epoch": 2.506550218340611, "percentage": 83.55, "elapsed_time": "22:49:02", "remaining_time": "4:29:30"} +{"current_steps": 5745, "total_steps": 6870, "loss": 0.2186, "lr": 3.9811570385716705e-06, "epoch": 2.5087336244541483, "percentage": 83.62, "elapsed_time": "22:50:14", "remaining_time": "4:28:19"} +{"current_steps": 5750, "total_steps": 6870, "loss": 0.2143, "lr": 3.9468379967917025e-06, "epoch": 2.5109170305676853, "percentage": 83.7, "elapsed_time": "22:51:25", "remaining_time": "4:27:07"} +{"current_steps": 5755, "total_steps": 6870, "loss": 0.2084, "lr": 3.912654835894564e-06, "epoch": 2.513100436681223, "percentage": 83.77, "elapsed_time": "22:52:37", "remaining_time": "4:25:56"} +{"current_steps": 5760, "total_steps": 6870, "loss": 0.2089, "lr": 3.878607776504489e-06, "epoch": 2.51528384279476, "percentage": 83.84, "elapsed_time": "22:53:48", "remaining_time": "4:24:44"} +{"current_steps": 5765, "total_steps": 6870, "loss": 0.2116, "lr": 3.844697038367314e-06, "epoch": 2.517467248908297, "percentage": 83.92, "elapsed_time": "22:55:00", "remaining_time": "4:23:33"} +{"current_steps": 5770, "total_steps": 6870, "loss": 0.2038, "lr": 3.8109228403490268e-06, "epoch": 2.519650655021834, "percentage": 83.99, "elapsed_time": "22:56:11", "remaining_time": "4:22:21"} +{"current_steps": 5775, "total_steps": 6870, "loss": 0.2116, "lr": 3.7772854004343494e-06, "epoch": 2.521834061135371, "percentage": 84.06, "elapsed_time": "22:57:23", "remaining_time": "4:21:10"} +{"current_steps": 5780, "total_steps": 6870, "loss": 0.2053, "lr": 3.7437849357253555e-06, "epoch": 2.524017467248908, "percentage": 84.13, "elapsed_time": "22:58:34", "remaining_time": "4:19:58"} +{"current_steps": 5785, "total_steps": 6870, "loss": 0.2114, "lr": 3.7104216624400502e-06, "epoch": 2.5262008733624453, "percentage": 84.21, "elapsed_time": "22:59:46", "remaining_time": "4:18:46"} +{"current_steps": 5790, "total_steps": 6870, "loss": 0.2054, "lr": 3.677195795910984e-06, "epoch": 2.5283842794759828, "percentage": 84.28, "elapsed_time": "23:00:58", "remaining_time": "4:17:35"} +{"current_steps": 5795, "total_steps": 6870, "loss": 0.2114, "lr": 3.6441075505838586e-06, "epoch": 2.53056768558952, "percentage": 84.35, "elapsed_time": "23:02:09", "remaining_time": "4:16:23"} +{"current_steps": 5800, "total_steps": 6870, "loss": 0.2107, "lr": 3.6111571400161387e-06, "epoch": 2.532751091703057, "percentage": 84.43, "elapsed_time": "23:03:21", "remaining_time": "4:15:12"} +{"current_steps": 5805, "total_steps": 6870, "loss": 0.2026, "lr": 3.578344776875689e-06, "epoch": 2.534934497816594, "percentage": 84.5, "elapsed_time": "23:04:32", "remaining_time": "4:14:00"} +{"current_steps": 5810, "total_steps": 6870, "loss": 0.209, "lr": 3.5456706729393774e-06, "epoch": 2.537117903930131, "percentage": 84.57, "elapsed_time": "23:05:44", "remaining_time": "4:12:49"} +{"current_steps": 5815, "total_steps": 6870, "loss": 0.2135, "lr": 3.5131350390917333e-06, "epoch": 2.539301310043668, "percentage": 84.64, "elapsed_time": "23:06:55", "remaining_time": "4:11:37"} +{"current_steps": 5820, "total_steps": 6870, "loss": 0.2203, "lr": 3.4807380853235704e-06, "epoch": 2.541484716157205, "percentage": 84.72, "elapsed_time": "23:08:07", "remaining_time": "4:10:26"} +{"current_steps": 5825, "total_steps": 6870, "loss": 0.2102, "lr": 3.4484800207306405e-06, "epoch": 2.5436681222707422, "percentage": 84.79, "elapsed_time": "23:09:18", "remaining_time": "4:09:14"} +{"current_steps": 5830, "total_steps": 6870, "loss": 0.2084, "lr": 3.4163610535122738e-06, "epoch": 2.5458515283842793, "percentage": 84.86, "elapsed_time": "23:10:30", "remaining_time": "4:08:02"} +{"current_steps": 5835, "total_steps": 6870, "loss": 0.2047, "lr": 3.384381390970048e-06, "epoch": 2.5480349344978164, "percentage": 84.93, "elapsed_time": "23:11:41", "remaining_time": "4:06:51"} +{"current_steps": 5840, "total_steps": 6870, "loss": 0.2024, "lr": 3.3525412395064375e-06, "epoch": 2.5502183406113534, "percentage": 85.01, "elapsed_time": "23:12:53", "remaining_time": "4:05:39"} +{"current_steps": 5845, "total_steps": 6870, "loss": 0.203, "lr": 3.32084080462349e-06, "epoch": 2.552401746724891, "percentage": 85.08, "elapsed_time": "23:14:05", "remaining_time": "4:04:28"} +{"current_steps": 5850, "total_steps": 6870, "loss": 0.2095, "lr": 3.2892802909214955e-06, "epoch": 2.554585152838428, "percentage": 85.15, "elapsed_time": "23:15:16", "remaining_time": "4:03:16"} +{"current_steps": 5855, "total_steps": 6870, "loss": 0.1952, "lr": 3.2578599020976707e-06, "epoch": 2.556768558951965, "percentage": 85.23, "elapsed_time": "23:16:28", "remaining_time": "4:02:05"} +{"current_steps": 5860, "total_steps": 6870, "loss": 0.2112, "lr": 3.2265798409448376e-06, "epoch": 2.558951965065502, "percentage": 85.3, "elapsed_time": "23:17:39", "remaining_time": "4:00:53"} +{"current_steps": 5865, "total_steps": 6870, "loss": 0.2182, "lr": 3.1954403093501227e-06, "epoch": 2.561135371179039, "percentage": 85.37, "elapsed_time": "23:18:51", "remaining_time": "3:59:42"} +{"current_steps": 5870, "total_steps": 6870, "loss": 0.2016, "lr": 3.1644415082936417e-06, "epoch": 2.5633187772925763, "percentage": 85.44, "elapsed_time": "23:20:03", "remaining_time": "3:58:30"} +{"current_steps": 5875, "total_steps": 6870, "loss": 0.1964, "lr": 3.1335836378472233e-06, "epoch": 2.565502183406114, "percentage": 85.52, "elapsed_time": "23:21:14", "remaining_time": "3:57:19"} +{"current_steps": 5880, "total_steps": 6870, "loss": 0.2076, "lr": 3.10286689717309e-06, "epoch": 2.567685589519651, "percentage": 85.59, "elapsed_time": "23:22:26", "remaining_time": "3:56:07"} +{"current_steps": 5885, "total_steps": 6870, "loss": 0.2109, "lr": 3.0722914845225963e-06, "epoch": 2.569868995633188, "percentage": 85.66, "elapsed_time": "23:23:37", "remaining_time": "3:54:55"} +{"current_steps": 5890, "total_steps": 6870, "loss": 0.1983, "lr": 3.04185759723494e-06, "epoch": 2.572052401746725, "percentage": 85.74, "elapsed_time": "23:24:49", "remaining_time": "3:53:44"} +{"current_steps": 5895, "total_steps": 6870, "loss": 0.2119, "lr": 3.0115654317358884e-06, "epoch": 2.574235807860262, "percentage": 85.81, "elapsed_time": "23:26:01", "remaining_time": "3:52:32"} +{"current_steps": 5900, "total_steps": 6870, "loss": 0.2121, "lr": 2.981415183536504e-06, "epoch": 2.576419213973799, "percentage": 85.88, "elapsed_time": "23:27:12", "remaining_time": "3:51:21"} +{"current_steps": 5905, "total_steps": 6870, "loss": 0.196, "lr": 2.9514070472318965e-06, "epoch": 2.578602620087336, "percentage": 85.95, "elapsed_time": "23:28:24", "remaining_time": "3:50:09"} +{"current_steps": 5910, "total_steps": 6870, "loss": 0.2007, "lr": 2.9215412164999594e-06, "epoch": 2.5807860262008733, "percentage": 86.03, "elapsed_time": "23:29:35", "remaining_time": "3:48:58"} +{"current_steps": 5915, "total_steps": 6870, "loss": 0.1918, "lr": 2.891817884100123e-06, "epoch": 2.5829694323144103, "percentage": 86.1, "elapsed_time": "23:30:47", "remaining_time": "3:47:46"} +{"current_steps": 5920, "total_steps": 6870, "loss": 0.2218, "lr": 2.862237241872101e-06, "epoch": 2.5851528384279474, "percentage": 86.17, "elapsed_time": "23:31:58", "remaining_time": "3:46:35"} +{"current_steps": 5925, "total_steps": 6870, "loss": 0.2243, "lr": 2.832799480734663e-06, "epoch": 2.5873362445414845, "percentage": 86.24, "elapsed_time": "23:33:10", "remaining_time": "3:45:23"} +{"current_steps": 5930, "total_steps": 6870, "loss": 0.2053, "lr": 2.8035047906844e-06, "epoch": 2.589519650655022, "percentage": 86.32, "elapsed_time": "23:34:22", "remaining_time": "3:44:11"} +{"current_steps": 5935, "total_steps": 6870, "loss": 0.2049, "lr": 2.774353360794493e-06, "epoch": 2.591703056768559, "percentage": 86.39, "elapsed_time": "23:35:33", "remaining_time": "3:43:00"} +{"current_steps": 5940, "total_steps": 6870, "loss": 0.2128, "lr": 2.7453453792134986e-06, "epoch": 2.593886462882096, "percentage": 86.46, "elapsed_time": "23:36:45", "remaining_time": "3:41:48"} +{"current_steps": 5945, "total_steps": 6870, "loss": 0.2036, "lr": 2.7164810331641366e-06, "epoch": 2.596069868995633, "percentage": 86.54, "elapsed_time": "23:37:56", "remaining_time": "3:40:37"} +{"current_steps": 5950, "total_steps": 6870, "loss": 0.2117, "lr": 2.687760508942064e-06, "epoch": 2.5982532751091703, "percentage": 86.61, "elapsed_time": "23:39:08", "remaining_time": "3:39:25"} +{"current_steps": 5955, "total_steps": 6870, "loss": 0.2144, "lr": 2.659183991914696e-06, "epoch": 2.6004366812227073, "percentage": 86.68, "elapsed_time": "23:40:19", "remaining_time": "3:38:14"} +{"current_steps": 5960, "total_steps": 6870, "loss": 0.2081, "lr": 2.6307516665200012e-06, "epoch": 2.6026200873362444, "percentage": 86.75, "elapsed_time": "23:41:31", "remaining_time": "3:37:02"} +{"current_steps": 5965, "total_steps": 6870, "loss": 0.2187, "lr": 2.602463716265302e-06, "epoch": 2.604803493449782, "percentage": 86.83, "elapsed_time": "23:42:42", "remaining_time": "3:35:51"} +{"current_steps": 5970, "total_steps": 6870, "loss": 0.2091, "lr": 2.5743203237261065e-06, "epoch": 2.606986899563319, "percentage": 86.9, "elapsed_time": "23:43:54", "remaining_time": "3:34:39"} +{"current_steps": 5975, "total_steps": 6870, "loss": 0.2178, "lr": 2.5463216705449155e-06, "epoch": 2.609170305676856, "percentage": 86.97, "elapsed_time": "23:45:06", "remaining_time": "3:33:28"} +{"current_steps": 5980, "total_steps": 6870, "loss": 0.2089, "lr": 2.5184679374300553e-06, "epoch": 2.611353711790393, "percentage": 87.05, "elapsed_time": "23:46:17", "remaining_time": "3:32:16"} +{"current_steps": 5985, "total_steps": 6870, "loss": 0.2166, "lr": 2.4907593041545163e-06, "epoch": 2.61353711790393, "percentage": 87.12, "elapsed_time": "23:47:29", "remaining_time": "3:31:04"} +{"current_steps": 5990, "total_steps": 6870, "loss": 0.1996, "lr": 2.4631959495547833e-06, "epoch": 2.6157205240174672, "percentage": 87.19, "elapsed_time": "23:48:40", "remaining_time": "3:29:53"} +{"current_steps": 5995, "total_steps": 6870, "loss": 0.1942, "lr": 2.4357780515297e-06, "epoch": 2.6179039301310043, "percentage": 87.26, "elapsed_time": "23:49:52", "remaining_time": "3:28:41"} +{"current_steps": 6000, "total_steps": 6870, "loss": 0.2024, "lr": 2.408505787039281e-06, "epoch": 2.6200873362445414, "percentage": 87.34, "elapsed_time": "23:51:04", "remaining_time": "3:27:30"} +{"current_steps": 6005, "total_steps": 6870, "loss": 0.2038, "lr": 2.381379332103623e-06, "epoch": 2.6222707423580784, "percentage": 87.41, "elapsed_time": "23:52:15", "remaining_time": "3:26:18"} +{"current_steps": 6010, "total_steps": 6870, "loss": 0.2005, "lr": 2.3543988618017237e-06, "epoch": 2.6244541484716155, "percentage": 87.48, "elapsed_time": "23:53:27", "remaining_time": "3:25:07"} +{"current_steps": 6015, "total_steps": 6870, "loss": 0.2004, "lr": 2.327564550270381e-06, "epoch": 2.6266375545851526, "percentage": 87.55, "elapsed_time": "23:54:38", "remaining_time": "3:23:55"} +{"current_steps": 6020, "total_steps": 6870, "loss": 0.1987, "lr": 2.3008765707030532e-06, "epoch": 2.62882096069869, "percentage": 87.63, "elapsed_time": "23:55:50", "remaining_time": "3:22:44"} +{"current_steps": 6025, "total_steps": 6870, "loss": 0.2161, "lr": 2.2743350953487424e-06, "epoch": 2.631004366812227, "percentage": 87.7, "elapsed_time": "23:57:01", "remaining_time": "3:21:32"} +{"current_steps": 6030, "total_steps": 6870, "loss": 0.1995, "lr": 2.2479402955108945e-06, "epoch": 2.6331877729257642, "percentage": 87.77, "elapsed_time": "23:58:13", "remaining_time": "3:20:20"} +{"current_steps": 6035, "total_steps": 6870, "loss": 0.207, "lr": 2.2216923415462802e-06, "epoch": 2.6353711790393013, "percentage": 87.85, "elapsed_time": "23:59:25", "remaining_time": "3:19:09"} +{"current_steps": 6040, "total_steps": 6870, "loss": 0.1994, "lr": 2.1955914028639e-06, "epoch": 2.6375545851528384, "percentage": 87.92, "elapsed_time": "1 day, 0:00:36", "remaining_time": "3:17:57"} +{"current_steps": 6045, "total_steps": 6870, "loss": 0.2099, "lr": 2.1696376479238947e-06, "epoch": 2.6397379912663754, "percentage": 87.99, "elapsed_time": "1 day, 0:01:48", "remaining_time": "3:16:46"} +{"current_steps": 6050, "total_steps": 6870, "loss": 0.2065, "lr": 2.1438312442364467e-06, "epoch": 2.641921397379913, "percentage": 88.06, "elapsed_time": "1 day, 0:02:59", "remaining_time": "3:15:34"} +{"current_steps": 6055, "total_steps": 6870, "loss": 0.201, "lr": 2.1181723583607177e-06, "epoch": 2.64410480349345, "percentage": 88.14, "elapsed_time": "1 day, 0:04:11", "remaining_time": "3:14:23"} +{"current_steps": 6060, "total_steps": 6870, "loss": 0.2075, "lr": 2.092661155903758e-06, "epoch": 2.646288209606987, "percentage": 88.21, "elapsed_time": "1 day, 0:05:22", "remaining_time": "3:13:11"} +{"current_steps": 6065, "total_steps": 6870, "loss": 0.1997, "lr": 2.067297801519441e-06, "epoch": 2.648471615720524, "percentage": 88.28, "elapsed_time": "1 day, 0:06:34", "remaining_time": "3:12:00"} +{"current_steps": 6070, "total_steps": 6870, "loss": 0.2066, "lr": 2.0420824589074073e-06, "epoch": 2.650655021834061, "percentage": 88.36, "elapsed_time": "1 day, 0:07:45", "remaining_time": "3:10:48"} +{"current_steps": 6075, "total_steps": 6870, "loss": 0.208, "lr": 2.0170152908119964e-06, "epoch": 2.6528384279475983, "percentage": 88.43, "elapsed_time": "1 day, 0:08:57", "remaining_time": "3:09:37"} +{"current_steps": 6080, "total_steps": 6870, "loss": 0.203, "lr": 1.9920964590212088e-06, "epoch": 2.6550218340611353, "percentage": 88.5, "elapsed_time": "1 day, 0:10:09", "remaining_time": "3:08:25"} +{"current_steps": 6085, "total_steps": 6870, "loss": 0.2048, "lr": 1.967326124365651e-06, "epoch": 2.6572052401746724, "percentage": 88.57, "elapsed_time": "1 day, 0:11:20", "remaining_time": "3:07:13"} +{"current_steps": 6090, "total_steps": 6870, "loss": 0.2055, "lr": 1.9427044467175054e-06, "epoch": 2.6593886462882095, "percentage": 88.65, "elapsed_time": "1 day, 0:12:32", "remaining_time": "3:06:02"} +{"current_steps": 6095, "total_steps": 6870, "loss": 0.1966, "lr": 1.9182315849894996e-06, "epoch": 2.6615720524017465, "percentage": 88.72, "elapsed_time": "1 day, 0:13:43", "remaining_time": "3:04:50"} +{"current_steps": 6100, "total_steps": 6870, "loss": 0.2134, "lr": 1.8939076971338592e-06, "epoch": 2.6637554585152836, "percentage": 88.79, "elapsed_time": "1 day, 0:14:55", "remaining_time": "3:03:39"} +{"current_steps": 6105, "total_steps": 6870, "loss": 0.2035, "lr": 1.8697329401413333e-06, "epoch": 2.665938864628821, "percentage": 88.86, "elapsed_time": "1 day, 0:16:07", "remaining_time": "3:02:27"} +{"current_steps": 6110, "total_steps": 6870, "loss": 0.2017, "lr": 1.8457074700401284e-06, "epoch": 2.668122270742358, "percentage": 88.94, "elapsed_time": "1 day, 0:17:18", "remaining_time": "3:01:16"} +{"current_steps": 6115, "total_steps": 6870, "loss": 0.2168, "lr": 1.821831441894939e-06, "epoch": 2.6703056768558953, "percentage": 89.01, "elapsed_time": "1 day, 0:18:30", "remaining_time": "3:00:04"} +{"current_steps": 6120, "total_steps": 6870, "loss": 0.205, "lr": 1.7981050098059364e-06, "epoch": 2.6724890829694323, "percentage": 89.08, "elapsed_time": "1 day, 0:19:41", "remaining_time": "2:58:53"} +{"current_steps": 6125, "total_steps": 6870, "loss": 0.1947, "lr": 1.7745283269077622e-06, "epoch": 2.6746724890829694, "percentage": 89.16, "elapsed_time": "1 day, 0:20:53", "remaining_time": "2:57:41"} +{"current_steps": 6130, "total_steps": 6870, "loss": 0.199, "lr": 1.751101545368558e-06, "epoch": 2.6768558951965065, "percentage": 89.23, "elapsed_time": "1 day, 0:22:04", "remaining_time": "2:56:29"} +{"current_steps": 6135, "total_steps": 6870, "loss": 0.2293, "lr": 1.7278248163889787e-06, "epoch": 2.6790393013100435, "percentage": 89.3, "elapsed_time": "1 day, 0:23:16", "remaining_time": "2:55:18"} +{"current_steps": 6140, "total_steps": 6870, "loss": 0.2006, "lr": 1.7046982902011954e-06, "epoch": 2.681222707423581, "percentage": 89.37, "elapsed_time": "1 day, 0:24:28", "remaining_time": "2:54:06"} +{"current_steps": 6145, "total_steps": 6870, "loss": 0.2167, "lr": 1.6817221160679603e-06, "epoch": 2.683406113537118, "percentage": 89.45, "elapsed_time": "1 day, 0:25:39", "remaining_time": "2:52:55"} +{"current_steps": 6150, "total_steps": 6870, "loss": 0.2138, "lr": 1.658896442281624e-06, "epoch": 2.685589519650655, "percentage": 89.52, "elapsed_time": "1 day, 0:26:51", "remaining_time": "2:51:43"} +{"current_steps": 6155, "total_steps": 6870, "loss": 0.1916, "lr": 1.636221416163175e-06, "epoch": 2.6877729257641922, "percentage": 89.59, "elapsed_time": "1 day, 0:28:02", "remaining_time": "2:50:32"} +{"current_steps": 5, "total_steps": 6870, "loss": 3.5144, "lr": 2.9112081513828244e-07, "epoch": 0.002183406113537118, "percentage": 0.07, "elapsed_time": "0:02:32", "remaining_time": "2 days, 10:04:24"} +{"current_steps": 10, "total_steps": 6870, "loss": 3.3991, "lr": 6.550218340611353e-07, "epoch": 0.004366812227074236, "percentage": 0.15, "elapsed_time": "0:04:58", "remaining_time": "2 days, 8:48:17"} +{"current_steps": 15, "total_steps": 6870, "loss": 3.4492, "lr": 1.0189228529839884e-06, "epoch": 0.006550218340611353, "percentage": 0.22, "elapsed_time": "0:07:23", "remaining_time": "2 days, 8:21:13"} +{"current_steps": 20, "total_steps": 6870, "loss": 3.3321, "lr": 1.3828238719068415e-06, "epoch": 0.008733624454148471, "percentage": 0.29, "elapsed_time": "0:09:49", "remaining_time": "2 days, 8:07:10"} +{"current_steps": 25, "total_steps": 6870, "loss": 3.0597, "lr": 1.7467248908296944e-06, "epoch": 0.010917030567685589, "percentage": 0.36, "elapsed_time": "0:12:15", "remaining_time": "2 days, 7:57:57"} +{"current_steps": 30, "total_steps": 6870, "loss": 2.9508, "lr": 2.1106259097525473e-06, "epoch": 0.013100436681222707, "percentage": 0.44, "elapsed_time": "0:14:41", "remaining_time": "2 days, 7:51:06"} +{"current_steps": 35, "total_steps": 6870, "loss": 2.612, "lr": 2.4745269286754004e-06, "epoch": 0.015283842794759825, "percentage": 0.51, "elapsed_time": "0:17:07", "remaining_time": "2 days, 7:45:17"} +{"current_steps": 40, "total_steps": 6870, "loss": 2.3435, "lr": 2.8384279475982535e-06, "epoch": 0.017467248908296942, "percentage": 0.58, "elapsed_time": "0:19:33", "remaining_time": "2 days, 7:40:22"} +{"current_steps": 45, "total_steps": 6870, "loss": 1.9812, "lr": 3.2023289665211066e-06, "epoch": 0.019650655021834062, "percentage": 0.66, "elapsed_time": "0:21:59", "remaining_time": "2 days, 7:36:00"} +{"current_steps": 50, "total_steps": 6870, "loss": 1.6557, "lr": 3.5662299854439593e-06, "epoch": 0.021834061135371178, "percentage": 0.73, "elapsed_time": "0:24:25", "remaining_time": "2 days, 7:31:58"} +{"current_steps": 55, "total_steps": 6870, "loss": 1.4334, "lr": 3.930131004366813e-06, "epoch": 0.024017467248908297, "percentage": 0.8, "elapsed_time": "0:26:52", "remaining_time": "2 days, 7:29:22"} +{"current_steps": 60, "total_steps": 6870, "loss": 1.2603, "lr": 4.2940320232896655e-06, "epoch": 0.026200873362445413, "percentage": 0.87, "elapsed_time": "0:29:18", "remaining_time": "2 days, 7:25:34"} +{"current_steps": 65, "total_steps": 6870, "loss": 1.128, "lr": 4.657933042212519e-06, "epoch": 0.028384279475982533, "percentage": 0.95, "elapsed_time": "0:31:43", "remaining_time": "2 days, 7:22:08"} +{"current_steps": 70, "total_steps": 6870, "loss": 1.0685, "lr": 5.021834061135372e-06, "epoch": 0.03056768558951965, "percentage": 1.02, "elapsed_time": "0:34:09", "remaining_time": "2 days, 7:18:11"} +{"current_steps": 75, "total_steps": 6870, "loss": 0.9865, "lr": 5.385735080058224e-06, "epoch": 0.03275109170305677, "percentage": 1.09, "elapsed_time": "0:36:35", "remaining_time": "2 days, 7:15:03"} +{"current_steps": 80, "total_steps": 6870, "loss": 0.8563, "lr": 5.749636098981077e-06, "epoch": 0.034934497816593885, "percentage": 1.16, "elapsed_time": "0:39:01", "remaining_time": "2 days, 7:11:54"} +{"current_steps": 85, "total_steps": 6870, "loss": 0.813, "lr": 6.11353711790393e-06, "epoch": 0.03711790393013101, "percentage": 1.24, "elapsed_time": "0:41:27", "remaining_time": "2 days, 7:09:00"} +{"current_steps": 90, "total_steps": 6870, "loss": 0.8189, "lr": 6.477438136826783e-06, "epoch": 0.039301310043668124, "percentage": 1.31, "elapsed_time": "0:43:53", "remaining_time": "2 days, 7:06:07"} +{"current_steps": 95, "total_steps": 6870, "loss": 0.7341, "lr": 6.841339155749636e-06, "epoch": 0.04148471615720524, "percentage": 1.38, "elapsed_time": "0:46:19", "remaining_time": "2 days, 7:03:10"} +{"current_steps": 100, "total_steps": 6870, "loss": 0.7264, "lr": 7.205240174672489e-06, "epoch": 0.043668122270742356, "percentage": 1.46, "elapsed_time": "0:48:44", "remaining_time": "2 days, 7:00:19"} +{"current_steps": 105, "total_steps": 6870, "loss": 0.7105, "lr": 7.569141193595343e-06, "epoch": 0.04585152838427948, "percentage": 1.53, "elapsed_time": "0:51:10", "remaining_time": "2 days, 6:57:32"} +{"current_steps": 110, "total_steps": 6870, "loss": 0.6883, "lr": 7.933042212518195e-06, "epoch": 0.048034934497816595, "percentage": 1.6, "elapsed_time": "0:53:36", "remaining_time": "2 days, 6:54:48"} +{"current_steps": 115, "total_steps": 6870, "loss": 0.6444, "lr": 8.296943231441049e-06, "epoch": 0.05021834061135371, "percentage": 1.67, "elapsed_time": "0:56:02", "remaining_time": "2 days, 6:52:01"} +{"current_steps": 120, "total_steps": 6870, "loss": 0.6228, "lr": 8.660844250363902e-06, "epoch": 0.05240174672489083, "percentage": 1.75, "elapsed_time": "0:58:28", "remaining_time": "2 days, 6:49:25"} +{"current_steps": 125, "total_steps": 6870, "loss": 0.6091, "lr": 9.024745269286754e-06, "epoch": 0.05458515283842795, "percentage": 1.82, "elapsed_time": "1:00:54", "remaining_time": "2 days, 6:46:45"} +{"current_steps": 130, "total_steps": 6870, "loss": 0.6049, "lr": 9.388646288209607e-06, "epoch": 0.056768558951965066, "percentage": 1.89, "elapsed_time": "1:03:20", "remaining_time": "2 days, 6:44:05"} +{"current_steps": 135, "total_steps": 6870, "loss": 0.6334, "lr": 9.75254730713246e-06, "epoch": 0.05895196506550218, "percentage": 1.97, "elapsed_time": "1:05:46", "remaining_time": "2 days, 6:41:28"} +{"current_steps": 140, "total_steps": 6870, "loss": 0.5484, "lr": 1.0116448326055314e-05, "epoch": 0.0611353711790393, "percentage": 2.04, "elapsed_time": "1:08:12", "remaining_time": "2 days, 6:38:50"} +{"current_steps": 145, "total_steps": 6870, "loss": 0.5777, "lr": 1.0480349344978165e-05, "epoch": 0.06331877729257641, "percentage": 2.11, "elapsed_time": "1:10:38", "remaining_time": "2 days, 6:36:13"} +{"current_steps": 150, "total_steps": 6870, "loss": 0.5427, "lr": 1.084425036390102e-05, "epoch": 0.06550218340611354, "percentage": 2.18, "elapsed_time": "1:13:04", "remaining_time": "2 days, 6:33:32"} +{"current_steps": 155, "total_steps": 6870, "loss": 0.5501, "lr": 1.1208151382823872e-05, "epoch": 0.06768558951965066, "percentage": 2.26, "elapsed_time": "1:15:30", "remaining_time": "2 days, 6:30:51"} +{"current_steps": 160, "total_steps": 6870, "loss": 0.5404, "lr": 1.1572052401746727e-05, "epoch": 0.06986899563318777, "percentage": 2.33, "elapsed_time": "1:17:55", "remaining_time": "2 days, 6:28:18"} +{"current_steps": 165, "total_steps": 6870, "loss": 0.5317, "lr": 1.1935953420669578e-05, "epoch": 0.07205240174672489, "percentage": 2.4, "elapsed_time": "1:20:21", "remaining_time": "2 days, 6:25:45"} +{"current_steps": 170, "total_steps": 6870, "loss": 0.5391, "lr": 1.2299854439592432e-05, "epoch": 0.07423580786026202, "percentage": 2.47, "elapsed_time": "1:22:47", "remaining_time": "2 days, 6:23:09"} +{"current_steps": 175, "total_steps": 6870, "loss": 0.5163, "lr": 1.2663755458515283e-05, "epoch": 0.07641921397379912, "percentage": 2.55, "elapsed_time": "1:25:13", "remaining_time": "2 days, 6:20:39"} +{"current_steps": 180, "total_steps": 6870, "loss": 0.5206, "lr": 1.3027656477438139e-05, "epoch": 0.07860262008733625, "percentage": 2.62, "elapsed_time": "1:27:39", "remaining_time": "2 days, 6:18:07"} +{"current_steps": 185, "total_steps": 6870, "loss": 0.51, "lr": 1.339155749636099e-05, "epoch": 0.08078602620087336, "percentage": 2.69, "elapsed_time": "1:30:05", "remaining_time": "2 days, 6:15:35"} +{"current_steps": 190, "total_steps": 6870, "loss": 0.4995, "lr": 1.3755458515283843e-05, "epoch": 0.08296943231441048, "percentage": 2.77, "elapsed_time": "1:32:31", "remaining_time": "2 days, 6:13:05"} +{"current_steps": 195, "total_steps": 6870, "loss": 0.4992, "lr": 1.4119359534206697e-05, "epoch": 0.0851528384279476, "percentage": 2.84, "elapsed_time": "1:34:57", "remaining_time": "2 days, 6:10:32"} +{"current_steps": 200, "total_steps": 6870, "loss": 0.5157, "lr": 1.448326055312955e-05, "epoch": 0.08733624454148471, "percentage": 2.91, "elapsed_time": "1:37:23", "remaining_time": "2 days, 6:08:02"} +{"current_steps": 205, "total_steps": 6870, "loss": 0.4884, "lr": 1.4847161572052404e-05, "epoch": 0.08951965065502183, "percentage": 2.98, "elapsed_time": "1:39:49", "remaining_time": "2 days, 6:05:33"} +{"current_steps": 210, "total_steps": 6870, "loss": 0.4837, "lr": 1.5211062590975255e-05, "epoch": 0.09170305676855896, "percentage": 3.06, "elapsed_time": "1:42:15", "remaining_time": "2 days, 6:03:04"} +{"current_steps": 215, "total_steps": 6870, "loss": 0.4614, "lr": 1.557496360989811e-05, "epoch": 0.09388646288209607, "percentage": 3.13, "elapsed_time": "1:44:41", "remaining_time": "2 days, 6:00:36"} +{"current_steps": 220, "total_steps": 6870, "loss": 0.5178, "lr": 1.5938864628820962e-05, "epoch": 0.09606986899563319, "percentage": 3.2, "elapsed_time": "1:47:07", "remaining_time": "2 days, 5:58:07"} +{"current_steps": 225, "total_steps": 6870, "loss": 0.4785, "lr": 1.6302765647743815e-05, "epoch": 0.0982532751091703, "percentage": 3.28, "elapsed_time": "1:49:33", "remaining_time": "2 days, 5:55:36"} +{"current_steps": 230, "total_steps": 6870, "loss": 0.4867, "lr": 1.6666666666666667e-05, "epoch": 0.10043668122270742, "percentage": 3.35, "elapsed_time": "1:51:59", "remaining_time": "2 days, 5:53:08"} +{"current_steps": 235, "total_steps": 6870, "loss": 0.4781, "lr": 1.703056768558952e-05, "epoch": 0.10262008733624454, "percentage": 3.42, "elapsed_time": "1:54:25", "remaining_time": "2 days, 5:50:34"} +{"current_steps": 240, "total_steps": 6870, "loss": 0.4732, "lr": 1.7394468704512373e-05, "epoch": 0.10480349344978165, "percentage": 3.49, "elapsed_time": "1:56:51", "remaining_time": "2 days, 5:48:00"} +{"current_steps": 245, "total_steps": 6870, "loss": 0.4651, "lr": 1.7758369723435225e-05, "epoch": 0.10698689956331878, "percentage": 3.57, "elapsed_time": "1:59:16", "remaining_time": "2 days, 5:45:28"} +{"current_steps": 250, "total_steps": 6870, "loss": 0.4585, "lr": 1.812227074235808e-05, "epoch": 0.1091703056768559, "percentage": 3.64, "elapsed_time": "2:01:42", "remaining_time": "2 days, 5:42:56"} +{"current_steps": 255, "total_steps": 6870, "loss": 0.4609, "lr": 1.848617176128093e-05, "epoch": 0.11135371179039301, "percentage": 3.71, "elapsed_time": "2:04:08", "remaining_time": "2 days, 5:40:24"} +{"current_steps": 260, "total_steps": 6870, "loss": 0.4549, "lr": 1.8850072780203784e-05, "epoch": 0.11353711790393013, "percentage": 3.78, "elapsed_time": "2:06:34", "remaining_time": "2 days, 5:37:51"} +{"current_steps": 265, "total_steps": 6870, "loss": 0.4783, "lr": 1.921397379912664e-05, "epoch": 0.11572052401746726, "percentage": 3.86, "elapsed_time": "2:09:00", "remaining_time": "2 days, 5:35:18"} +{"current_steps": 270, "total_steps": 6870, "loss": 0.455, "lr": 1.9577874818049492e-05, "epoch": 0.11790393013100436, "percentage": 3.93, "elapsed_time": "2:11:25", "remaining_time": "2 days, 5:32:47"} +{"current_steps": 275, "total_steps": 6870, "loss": 0.4479, "lr": 1.9941775836972345e-05, "epoch": 0.12008733624454149, "percentage": 4.0, "elapsed_time": "2:13:51", "remaining_time": "2 days, 5:30:19"} +{"current_steps": 280, "total_steps": 6870, "loss": 0.456, "lr": 2.0305676855895198e-05, "epoch": 0.1222707423580786, "percentage": 4.08, "elapsed_time": "2:16:17", "remaining_time": "2 days, 5:27:52"} +{"current_steps": 285, "total_steps": 6870, "loss": 0.4318, "lr": 2.066957787481805e-05, "epoch": 0.12445414847161572, "percentage": 4.15, "elapsed_time": "2:18:43", "remaining_time": "2 days, 5:25:24"} +{"current_steps": 290, "total_steps": 6870, "loss": 0.4494, "lr": 2.1033478893740903e-05, "epoch": 0.12663755458515283, "percentage": 4.22, "elapsed_time": "2:21:09", "remaining_time": "2 days, 5:22:58"} +{"current_steps": 295, "total_steps": 6870, "loss": 0.4604, "lr": 2.1397379912663756e-05, "epoch": 0.12882096069868995, "percentage": 4.29, "elapsed_time": "2:23:35", "remaining_time": "2 days, 5:20:32"} +{"current_steps": 300, "total_steps": 6870, "loss": 0.4529, "lr": 2.176128093158661e-05, "epoch": 0.13100436681222707, "percentage": 4.37, "elapsed_time": "2:26:01", "remaining_time": "2 days, 5:18:06"} +{"current_steps": 305, "total_steps": 6870, "loss": 0.4462, "lr": 2.2125181950509464e-05, "epoch": 0.1331877729257642, "percentage": 4.44, "elapsed_time": "2:28:27", "remaining_time": "2 days, 5:15:40"} +{"current_steps": 310, "total_steps": 6870, "loss": 0.4353, "lr": 2.2489082969432314e-05, "epoch": 0.13537117903930132, "percentage": 4.51, "elapsed_time": "2:30:53", "remaining_time": "2 days, 5:13:13"} +{"current_steps": 315, "total_steps": 6870, "loss": 0.4425, "lr": 2.2852983988355166e-05, "epoch": 0.13755458515283842, "percentage": 4.59, "elapsed_time": "2:33:19", "remaining_time": "2 days, 5:10:45"} +{"current_steps": 320, "total_steps": 6870, "loss": 0.4375, "lr": 2.3216885007278022e-05, "epoch": 0.13973799126637554, "percentage": 4.66, "elapsed_time": "2:35:45", "remaining_time": "2 days, 5:08:17"} +{"current_steps": 325, "total_steps": 6870, "loss": 0.4539, "lr": 2.3580786026200875e-05, "epoch": 0.14192139737991266, "percentage": 4.73, "elapsed_time": "2:38:11", "remaining_time": "2 days, 5:05:50"} +{"current_steps": 330, "total_steps": 6870, "loss": 0.4347, "lr": 2.3944687045123728e-05, "epoch": 0.14410480349344978, "percentage": 4.8, "elapsed_time": "2:40:37", "remaining_time": "2 days, 5:03:23"} +{"current_steps": 335, "total_steps": 6870, "loss": 0.4149, "lr": 2.430858806404658e-05, "epoch": 0.1462882096069869, "percentage": 4.88, "elapsed_time": "2:43:03", "remaining_time": "2 days, 5:00:55"} +{"current_steps": 340, "total_steps": 6870, "loss": 0.422, "lr": 2.4672489082969433e-05, "epoch": 0.14847161572052403, "percentage": 4.95, "elapsed_time": "2:45:29", "remaining_time": "2 days, 4:58:20"} +{"current_steps": 345, "total_steps": 6870, "loss": 0.4418, "lr": 2.503639010189229e-05, "epoch": 0.15065502183406113, "percentage": 5.02, "elapsed_time": "2:47:55", "remaining_time": "2 days, 4:55:54"} +{"current_steps": 350, "total_steps": 6870, "loss": 0.4404, "lr": 2.540029112081514e-05, "epoch": 0.15283842794759825, "percentage": 5.09, "elapsed_time": "2:50:21", "remaining_time": "2 days, 4:53:28"} +{"current_steps": 355, "total_steps": 6870, "loss": 0.4263, "lr": 2.576419213973799e-05, "epoch": 0.15502183406113537, "percentage": 5.17, "elapsed_time": "2:52:47", "remaining_time": "2 days, 4:51:01"} +{"current_steps": 360, "total_steps": 6870, "loss": 0.4299, "lr": 2.6128093158660844e-05, "epoch": 0.1572052401746725, "percentage": 5.24, "elapsed_time": "2:55:13", "remaining_time": "2 days, 4:48:31"} +{"current_steps": 365, "total_steps": 6870, "loss": 0.4196, "lr": 2.64919941775837e-05, "epoch": 0.15938864628820962, "percentage": 5.31, "elapsed_time": "2:57:39", "remaining_time": "2 days, 4:46:05"} +{"current_steps": 370, "total_steps": 6870, "loss": 0.4174, "lr": 2.6855895196506553e-05, "epoch": 0.1615720524017467, "percentage": 5.39, "elapsed_time": "3:00:05", "remaining_time": "2 days, 4:43:40"} +{"current_steps": 375, "total_steps": 6870, "loss": 0.3993, "lr": 2.7219796215429405e-05, "epoch": 0.16375545851528384, "percentage": 5.46, "elapsed_time": "3:02:31", "remaining_time": "2 days, 4:41:16"} +{"current_steps": 380, "total_steps": 6870, "loss": 0.4345, "lr": 2.7583697234352258e-05, "epoch": 0.16593886462882096, "percentage": 5.53, "elapsed_time": "3:04:57", "remaining_time": "2 days, 4:38:49"} +{"current_steps": 385, "total_steps": 6870, "loss": 0.4441, "lr": 2.794759825327511e-05, "epoch": 0.16812227074235808, "percentage": 5.6, "elapsed_time": "3:07:23", "remaining_time": "2 days, 4:36:22"} +{"current_steps": 390, "total_steps": 6870, "loss": 0.4029, "lr": 2.831149927219796e-05, "epoch": 0.1703056768558952, "percentage": 5.68, "elapsed_time": "3:09:49", "remaining_time": "2 days, 4:33:55"} +{"current_steps": 395, "total_steps": 6870, "loss": 0.4342, "lr": 2.867540029112082e-05, "epoch": 0.17248908296943233, "percentage": 5.75, "elapsed_time": "3:12:15", "remaining_time": "2 days, 4:31:28"} +{"current_steps": 400, "total_steps": 6870, "loss": 0.4044, "lr": 2.9039301310043672e-05, "epoch": 0.17467248908296942, "percentage": 5.82, "elapsed_time": "3:14:41", "remaining_time": "2 days, 4:29:01"} +{"current_steps": 405, "total_steps": 6870, "loss": 0.4114, "lr": 2.940320232896652e-05, "epoch": 0.17685589519650655, "percentage": 5.9, "elapsed_time": "3:17:07", "remaining_time": "2 days, 4:26:37"} +{"current_steps": 410, "total_steps": 6870, "loss": 0.4124, "lr": 2.9767103347889374e-05, "epoch": 0.17903930131004367, "percentage": 5.97, "elapsed_time": "3:19:33", "remaining_time": "2 days, 4:24:11"} +{"current_steps": 415, "total_steps": 6870, "loss": 0.4096, "lr": 3.0131004366812227e-05, "epoch": 0.1812227074235808, "percentage": 6.04, "elapsed_time": "3:21:59", "remaining_time": "2 days, 4:21:44"} +{"current_steps": 420, "total_steps": 6870, "loss": 0.4019, "lr": 3.0494905385735083e-05, "epoch": 0.18340611353711792, "percentage": 6.11, "elapsed_time": "3:24:25", "remaining_time": "2 days, 4:19:19"} +{"current_steps": 425, "total_steps": 6870, "loss": 0.4125, "lr": 3.0858806404657935e-05, "epoch": 0.185589519650655, "percentage": 6.19, "elapsed_time": "3:26:51", "remaining_time": "2 days, 4:16:52"} +{"current_steps": 430, "total_steps": 6870, "loss": 0.3979, "lr": 3.122270742358079e-05, "epoch": 0.18777292576419213, "percentage": 6.26, "elapsed_time": "3:29:17", "remaining_time": "2 days, 4:14:26"} +{"current_steps": 435, "total_steps": 6870, "loss": 0.411, "lr": 3.158660844250364e-05, "epoch": 0.18995633187772926, "percentage": 6.33, "elapsed_time": "3:31:43", "remaining_time": "2 days, 4:11:59"} +{"current_steps": 440, "total_steps": 6870, "loss": 0.4001, "lr": 3.1950509461426493e-05, "epoch": 0.19213973799126638, "percentage": 6.4, "elapsed_time": "3:34:09", "remaining_time": "2 days, 4:09:34"} +{"current_steps": 445, "total_steps": 6870, "loss": 0.3982, "lr": 3.2314410480349346e-05, "epoch": 0.1943231441048035, "percentage": 6.48, "elapsed_time": "3:36:35", "remaining_time": "2 days, 4:07:08"} +{"current_steps": 450, "total_steps": 6870, "loss": 0.4129, "lr": 3.26783114992722e-05, "epoch": 0.1965065502183406, "percentage": 6.55, "elapsed_time": "3:39:01", "remaining_time": "2 days, 4:04:42"} +{"current_steps": 455, "total_steps": 6870, "loss": 0.4173, "lr": 3.304221251819505e-05, "epoch": 0.19868995633187772, "percentage": 6.62, "elapsed_time": "3:41:27", "remaining_time": "2 days, 4:02:17"} +{"current_steps": 460, "total_steps": 6870, "loss": 0.3947, "lr": 3.3406113537117904e-05, "epoch": 0.20087336244541484, "percentage": 6.7, "elapsed_time": "3:43:53", "remaining_time": "2 days, 3:59:50"} +{"current_steps": 465, "total_steps": 6870, "loss": 0.4066, "lr": 3.377001455604076e-05, "epoch": 0.20305676855895197, "percentage": 6.77, "elapsed_time": "3:46:19", "remaining_time": "2 days, 3:57:24"} +{"current_steps": 470, "total_steps": 6870, "loss": 0.4054, "lr": 3.413391557496361e-05, "epoch": 0.2052401746724891, "percentage": 6.84, "elapsed_time": "3:48:45", "remaining_time": "2 days, 3:54:57"} +{"current_steps": 475, "total_steps": 6870, "loss": 0.3923, "lr": 3.449781659388647e-05, "epoch": 0.2074235807860262, "percentage": 6.91, "elapsed_time": "3:51:11", "remaining_time": "2 days, 3:52:32"} +{"current_steps": 480, "total_steps": 6870, "loss": 0.4068, "lr": 3.486171761280932e-05, "epoch": 0.2096069868995633, "percentage": 6.99, "elapsed_time": "3:53:37", "remaining_time": "2 days, 3:50:06"} +{"current_steps": 485, "total_steps": 6870, "loss": 0.4002, "lr": 3.5225618631732174e-05, "epoch": 0.21179039301310043, "percentage": 7.06, "elapsed_time": "3:56:03", "remaining_time": "2 days, 3:47:39"} +{"current_steps": 490, "total_steps": 6870, "loss": 0.4068, "lr": 3.558951965065502e-05, "epoch": 0.21397379912663755, "percentage": 7.13, "elapsed_time": "3:58:29", "remaining_time": "2 days, 3:45:12"} +{"current_steps": 495, "total_steps": 6870, "loss": 0.4081, "lr": 3.595342066957787e-05, "epoch": 0.21615720524017468, "percentage": 7.21, "elapsed_time": "4:00:55", "remaining_time": "2 days, 3:42:47"} +{"current_steps": 500, "total_steps": 6870, "loss": 0.4053, "lr": 3.6317321688500726e-05, "epoch": 0.2183406113537118, "percentage": 7.28, "elapsed_time": "4:03:21", "remaining_time": "2 days, 3:40:21"} +{"current_steps": 505, "total_steps": 6870, "loss": 0.391, "lr": 3.6681222707423585e-05, "epoch": 0.2205240174672489, "percentage": 7.35, "elapsed_time": "4:05:47", "remaining_time": "2 days, 3:37:55"} +{"current_steps": 510, "total_steps": 6870, "loss": 0.4061, "lr": 3.704512372634644e-05, "epoch": 0.22270742358078602, "percentage": 7.42, "elapsed_time": "4:08:13", "remaining_time": "2 days, 3:35:31"} +{"current_steps": 515, "total_steps": 6870, "loss": 0.4107, "lr": 3.740902474526929e-05, "epoch": 0.22489082969432314, "percentage": 7.5, "elapsed_time": "4:10:39", "remaining_time": "2 days, 3:33:05"} +{"current_steps": 520, "total_steps": 6870, "loss": 0.3998, "lr": 3.777292576419214e-05, "epoch": 0.22707423580786026, "percentage": 7.57, "elapsed_time": "4:13:05", "remaining_time": "2 days, 3:30:36"} +{"current_steps": 525, "total_steps": 6870, "loss": 0.4005, "lr": 3.8136826783114996e-05, "epoch": 0.2292576419213974, "percentage": 7.64, "elapsed_time": "4:15:31", "remaining_time": "2 days, 3:28:08"} +{"current_steps": 530, "total_steps": 6870, "loss": 0.4098, "lr": 3.850072780203784e-05, "epoch": 0.2314410480349345, "percentage": 7.71, "elapsed_time": "4:17:57", "remaining_time": "2 days, 3:25:43"} +{"current_steps": 535, "total_steps": 6870, "loss": 0.3965, "lr": 3.88646288209607e-05, "epoch": 0.2336244541484716, "percentage": 7.79, "elapsed_time": "4:20:23", "remaining_time": "2 days, 3:23:16"} +{"current_steps": 540, "total_steps": 6870, "loss": 0.3962, "lr": 3.9228529839883554e-05, "epoch": 0.23580786026200873, "percentage": 7.86, "elapsed_time": "4:22:49", "remaining_time": "2 days, 3:20:48"} +{"current_steps": 545, "total_steps": 6870, "loss": 0.3976, "lr": 3.9592430858806406e-05, "epoch": 0.23799126637554585, "percentage": 7.93, "elapsed_time": "4:25:14", "remaining_time": "2 days, 3:18:19"} +{"current_steps": 550, "total_steps": 6870, "loss": 0.3932, "lr": 3.995633187772926e-05, "epoch": 0.24017467248908297, "percentage": 8.01, "elapsed_time": "4:27:40", "remaining_time": "2 days, 3:15:52"} +{"current_steps": 555, "total_steps": 6870, "loss": 0.4163, "lr": 4.032023289665211e-05, "epoch": 0.2423580786026201, "percentage": 8.08, "elapsed_time": "4:30:06", "remaining_time": "2 days, 3:13:25"} +{"current_steps": 560, "total_steps": 6870, "loss": 0.3885, "lr": 4.068413391557497e-05, "epoch": 0.2445414847161572, "percentage": 8.15, "elapsed_time": "4:32:32", "remaining_time": "2 days, 3:10:59"} +{"current_steps": 565, "total_steps": 6870, "loss": 0.3979, "lr": 4.104803493449782e-05, "epoch": 0.24672489082969432, "percentage": 8.22, "elapsed_time": "4:34:58", "remaining_time": "2 days, 3:08:33"} +{"current_steps": 570, "total_steps": 6870, "loss": 0.4052, "lr": 4.141193595342067e-05, "epoch": 0.24890829694323144, "percentage": 8.3, "elapsed_time": "4:37:24", "remaining_time": "2 days, 3:06:06"} +{"current_steps": 575, "total_steps": 6870, "loss": 0.3931, "lr": 4.177583697234352e-05, "epoch": 0.25109170305676853, "percentage": 8.37, "elapsed_time": "4:39:50", "remaining_time": "2 days, 3:03:40"} +{"current_steps": 580, "total_steps": 6870, "loss": 0.4199, "lr": 4.2139737991266375e-05, "epoch": 0.25327510917030566, "percentage": 8.44, "elapsed_time": "4:42:16", "remaining_time": "2 days, 3:01:13"} +{"current_steps": 585, "total_steps": 6870, "loss": 0.3817, "lr": 4.250363901018923e-05, "epoch": 0.2554585152838428, "percentage": 8.52, "elapsed_time": "4:44:42", "remaining_time": "2 days, 2:58:47"} +{"current_steps": 590, "total_steps": 6870, "loss": 0.4146, "lr": 4.286754002911209e-05, "epoch": 0.2576419213973799, "percentage": 8.59, "elapsed_time": "4:47:08", "remaining_time": "2 days, 2:56:20"} +{"current_steps": 595, "total_steps": 6870, "loss": 0.3894, "lr": 4.323144104803494e-05, "epoch": 0.259825327510917, "percentage": 8.66, "elapsed_time": "4:49:34", "remaining_time": "2 days, 2:53:52"} +{"current_steps": 600, "total_steps": 6870, "loss": 0.4099, "lr": 4.359534206695779e-05, "epoch": 0.26200873362445415, "percentage": 8.73, "elapsed_time": "4:52:00", "remaining_time": "2 days, 2:51:26"} +{"current_steps": 605, "total_steps": 6870, "loss": 0.3937, "lr": 4.395924308588064e-05, "epoch": 0.26419213973799127, "percentage": 8.81, "elapsed_time": "4:54:26", "remaining_time": "2 days, 2:49:00"} +{"current_steps": 610, "total_steps": 6870, "loss": 0.4041, "lr": 4.432314410480349e-05, "epoch": 0.2663755458515284, "percentage": 8.88, "elapsed_time": "4:56:52", "remaining_time": "2 days, 2:46:33"} +{"current_steps": 615, "total_steps": 6870, "loss": 0.3923, "lr": 4.468704512372635e-05, "epoch": 0.2685589519650655, "percentage": 8.95, "elapsed_time": "4:59:18", "remaining_time": "2 days, 2:44:07"} +{"current_steps": 620, "total_steps": 6870, "loss": 0.3933, "lr": 4.50509461426492e-05, "epoch": 0.27074235807860264, "percentage": 9.02, "elapsed_time": "5:01:43", "remaining_time": "2 days, 2:41:37"} +{"current_steps": 625, "total_steps": 6870, "loss": 0.3887, "lr": 4.5414847161572056e-05, "epoch": 0.27292576419213976, "percentage": 9.1, "elapsed_time": "5:04:09", "remaining_time": "2 days, 2:39:10"} +{"current_steps": 630, "total_steps": 6870, "loss": 0.3767, "lr": 4.577874818049491e-05, "epoch": 0.27510917030567683, "percentage": 9.17, "elapsed_time": "5:06:35", "remaining_time": "2 days, 2:36:44"} +{"current_steps": 635, "total_steps": 6870, "loss": 0.3996, "lr": 4.614264919941776e-05, "epoch": 0.27729257641921395, "percentage": 9.24, "elapsed_time": "5:09:01", "remaining_time": "2 days, 2:34:17"} +{"current_steps": 640, "total_steps": 6870, "loss": 0.3966, "lr": 4.6506550218340614e-05, "epoch": 0.2794759825327511, "percentage": 9.32, "elapsed_time": "5:11:27", "remaining_time": "2 days, 2:31:51"} +{"current_steps": 645, "total_steps": 6870, "loss": 0.4006, "lr": 4.687045123726347e-05, "epoch": 0.2816593886462882, "percentage": 9.39, "elapsed_time": "5:13:53", "remaining_time": "2 days, 2:29:25"} +{"current_steps": 650, "total_steps": 6870, "loss": 0.3923, "lr": 4.723435225618632e-05, "epoch": 0.2838427947598253, "percentage": 9.46, "elapsed_time": "5:16:19", "remaining_time": "2 days, 2:26:59"} +{"current_steps": 655, "total_steps": 6870, "loss": 0.402, "lr": 4.759825327510917e-05, "epoch": 0.28602620087336245, "percentage": 9.53, "elapsed_time": "5:18:45", "remaining_time": "2 days, 2:24:34"} +{"current_steps": 660, "total_steps": 6870, "loss": 0.4099, "lr": 4.7962154294032025e-05, "epoch": 0.28820960698689957, "percentage": 9.61, "elapsed_time": "5:21:11", "remaining_time": "2 days, 2:22:08"} +{"current_steps": 665, "total_steps": 6870, "loss": 0.3886, "lr": 4.832605531295488e-05, "epoch": 0.2903930131004367, "percentage": 9.68, "elapsed_time": "5:23:37", "remaining_time": "2 days, 2:19:42"} +{"current_steps": 670, "total_steps": 6870, "loss": 0.3706, "lr": 4.868995633187774e-05, "epoch": 0.2925764192139738, "percentage": 9.75, "elapsed_time": "5:26:03", "remaining_time": "2 days, 2:17:16"} +{"current_steps": 675, "total_steps": 6870, "loss": 0.4012, "lr": 4.905385735080058e-05, "epoch": 0.29475982532751094, "percentage": 9.83, "elapsed_time": "5:28:29", "remaining_time": "2 days, 2:14:50"} +{"current_steps": 680, "total_steps": 6870, "loss": 0.3842, "lr": 4.9417758369723435e-05, "epoch": 0.29694323144104806, "percentage": 9.9, "elapsed_time": "5:30:55", "remaining_time": "2 days, 2:12:23"} +{"current_steps": 685, "total_steps": 6870, "loss": 0.3855, "lr": 4.978165938864629e-05, "epoch": 0.29912663755458513, "percentage": 9.97, "elapsed_time": "5:33:21", "remaining_time": "2 days, 2:09:57"} +{"current_steps": 690, "total_steps": 6870, "loss": 0.3705, "lr": 4.999998709163553e-05, "epoch": 0.30131004366812225, "percentage": 10.04, "elapsed_time": "5:35:47", "remaining_time": "2 days, 2:07:31"} +{"current_steps": 695, "total_steps": 6870, "loss": 0.3828, "lr": 4.999984187268831e-05, "epoch": 0.3034934497816594, "percentage": 10.12, "elapsed_time": "5:38:13", "remaining_time": "2 days, 2:05:05"} +{"current_steps": 700, "total_steps": 6870, "loss": 0.3763, "lr": 4.9999535300278666e-05, "epoch": 0.3056768558951965, "percentage": 10.19, "elapsed_time": "5:40:39", "remaining_time": "2 days, 2:02:40"} +{"current_steps": 705, "total_steps": 6870, "loss": 0.3928, "lr": 4.999906737638528e-05, "epoch": 0.3078602620087336, "percentage": 10.26, "elapsed_time": "5:43:05", "remaining_time": "2 days, 2:00:14"} +{"current_steps": 710, "total_steps": 6870, "loss": 0.3926, "lr": 4.9998438104028226e-05, "epoch": 0.31004366812227074, "percentage": 10.33, "elapsed_time": "5:45:31", "remaining_time": "2 days, 1:57:48"} +{"current_steps": 715, "total_steps": 6870, "loss": 0.385, "lr": 4.999764748726891e-05, "epoch": 0.31222707423580787, "percentage": 10.41, "elapsed_time": "5:47:57", "remaining_time": "2 days, 1:55:22"} +{"current_steps": 720, "total_steps": 6870, "loss": 0.4175, "lr": 4.9996695531210135e-05, "epoch": 0.314410480349345, "percentage": 10.48, "elapsed_time": "5:50:23", "remaining_time": "2 days, 1:52:57"} +{"current_steps": 725, "total_steps": 6870, "loss": 0.4076, "lr": 4.9995582241996e-05, "epoch": 0.3165938864628821, "percentage": 10.55, "elapsed_time": "5:52:49", "remaining_time": "2 days, 1:50:32"} +{"current_steps": 730, "total_steps": 6870, "loss": 0.3912, "lr": 4.9994307626811865e-05, "epoch": 0.31877729257641924, "percentage": 10.63, "elapsed_time": "5:55:15", "remaining_time": "2 days, 1:48:07"} +{"current_steps": 735, "total_steps": 6870, "loss": 0.3913, "lr": 4.999287169388433e-05, "epoch": 0.32096069868995636, "percentage": 10.7, "elapsed_time": "5:57:42", "remaining_time": "2 days, 1:45:42"} +{"current_steps": 740, "total_steps": 6870, "loss": 0.4078, "lr": 4.9991274452481164e-05, "epoch": 0.3231441048034934, "percentage": 10.77, "elapsed_time": "6:00:08", "remaining_time": "2 days, 1:43:18"} +{"current_steps": 745, "total_steps": 6870, "loss": 0.3957, "lr": 4.9989515912911234e-05, "epoch": 0.32532751091703055, "percentage": 10.84, "elapsed_time": "6:02:34", "remaining_time": "2 days, 1:40:53"} +{"current_steps": 750, "total_steps": 6870, "loss": 0.4049, "lr": 4.998759608652449e-05, "epoch": 0.32751091703056767, "percentage": 10.92, "elapsed_time": "6:05:00", "remaining_time": "2 days, 1:38:28"} +{"current_steps": 755, "total_steps": 6870, "loss": 0.3929, "lr": 4.998551498571182e-05, "epoch": 0.3296943231441048, "percentage": 10.99, "elapsed_time": "6:07:26", "remaining_time": "2 days, 1:36:02"} +{"current_steps": 760, "total_steps": 6870, "loss": 0.3823, "lr": 4.9983272623905045e-05, "epoch": 0.3318777292576419, "percentage": 11.06, "elapsed_time": "6:09:52", "remaining_time": "2 days, 1:33:37"} +{"current_steps": 765, "total_steps": 6870, "loss": 0.3761, "lr": 4.9980869015576736e-05, "epoch": 0.33406113537117904, "percentage": 11.14, "elapsed_time": "6:12:18", "remaining_time": "2 days, 1:31:11"} +{"current_steps": 770, "total_steps": 6870, "loss": 0.3867, "lr": 4.997830417624024e-05, "epoch": 0.33624454148471616, "percentage": 11.21, "elapsed_time": "6:14:44", "remaining_time": "2 days, 1:28:46"} +{"current_steps": 775, "total_steps": 6870, "loss": 0.3953, "lr": 4.9975578122449485e-05, "epoch": 0.3384279475982533, "percentage": 11.28, "elapsed_time": "6:17:10", "remaining_time": "2 days, 1:26:20"} +{"current_steps": 780, "total_steps": 6870, "loss": 0.3769, "lr": 4.99726908717989e-05, "epoch": 0.3406113537117904, "percentage": 11.35, "elapsed_time": "6:19:36", "remaining_time": "2 days, 1:23:54"} +{"current_steps": 785, "total_steps": 6870, "loss": 0.3776, "lr": 4.9969642442923325e-05, "epoch": 0.34279475982532753, "percentage": 11.43, "elapsed_time": "6:22:02", "remaining_time": "2 days, 1:21:29"} +{"current_steps": 790, "total_steps": 6870, "loss": 0.3912, "lr": 4.996643285549787e-05, "epoch": 0.34497816593886466, "percentage": 11.5, "elapsed_time": "6:24:29", "remaining_time": "2 days, 1:19:03"} +{"current_steps": 795, "total_steps": 6870, "loss": 0.374, "lr": 4.996306213023778e-05, "epoch": 0.3471615720524017, "percentage": 11.57, "elapsed_time": "6:26:55", "remaining_time": "2 days, 1:16:38"} +{"current_steps": 800, "total_steps": 6870, "loss": 0.3908, "lr": 4.995953028889833e-05, "epoch": 0.34934497816593885, "percentage": 11.64, "elapsed_time": "6:29:21", "remaining_time": "2 days, 1:14:12"} +{"current_steps": 805, "total_steps": 6870, "loss": 0.3804, "lr": 4.995583735427465e-05, "epoch": 0.35152838427947597, "percentage": 11.72, "elapsed_time": "6:31:47", "remaining_time": "2 days, 1:11:47"} +{"current_steps": 810, "total_steps": 6870, "loss": 0.3787, "lr": 4.9951983350201604e-05, "epoch": 0.3537117903930131, "percentage": 11.79, "elapsed_time": "6:34:13", "remaining_time": "2 days, 1:09:21"} +{"current_steps": 815, "total_steps": 6870, "loss": 0.3681, "lr": 4.9947968301553625e-05, "epoch": 0.3558951965065502, "percentage": 11.86, "elapsed_time": "6:36:39", "remaining_time": "2 days, 1:06:56"} +{"current_steps": 820, "total_steps": 6870, "loss": 0.3632, "lr": 4.994379223424456e-05, "epoch": 0.35807860262008734, "percentage": 11.94, "elapsed_time": "6:39:05", "remaining_time": "2 days, 1:04:31"} +{"current_steps": 825, "total_steps": 6870, "loss": 0.3729, "lr": 4.99394551752275e-05, "epoch": 0.36026200873362446, "percentage": 12.01, "elapsed_time": "6:41:31", "remaining_time": "2 days, 1:02:05"} +{"current_steps": 830, "total_steps": 6870, "loss": 0.366, "lr": 4.9934957152494585e-05, "epoch": 0.3624454148471616, "percentage": 12.08, "elapsed_time": "6:43:57", "remaining_time": "2 days, 0:59:40"} +{"current_steps": 835, "total_steps": 6870, "loss": 0.3641, "lr": 4.993029819507687e-05, "epoch": 0.3646288209606987, "percentage": 12.15, "elapsed_time": "6:46:23", "remaining_time": "2 days, 0:57:15"} +{"current_steps": 840, "total_steps": 6870, "loss": 0.3797, "lr": 4.9925478333044114e-05, "epoch": 0.36681222707423583, "percentage": 12.23, "elapsed_time": "6:48:49", "remaining_time": "2 days, 0:54:49"} +{"current_steps": 845, "total_steps": 6870, "loss": 0.3776, "lr": 4.992049759750455e-05, "epoch": 0.36899563318777295, "percentage": 12.3, "elapsed_time": "6:51:16", "remaining_time": "2 days, 0:52:24"} +{"current_steps": 850, "total_steps": 6870, "loss": 0.3723, "lr": 4.9915356020604754e-05, "epoch": 0.37117903930131, "percentage": 12.37, "elapsed_time": "6:53:42", "remaining_time": "2 days, 0:49:59"} +{"current_steps": 855, "total_steps": 6870, "loss": 0.3918, "lr": 4.9910053635529376e-05, "epoch": 0.37336244541484714, "percentage": 12.45, "elapsed_time": "6:56:08", "remaining_time": "2 days, 0:47:33"} +{"current_steps": 860, "total_steps": 6870, "loss": 0.3809, "lr": 4.9904590476500955e-05, "epoch": 0.37554585152838427, "percentage": 12.52, "elapsed_time": "6:58:34", "remaining_time": "2 days, 0:45:08"} +{"current_steps": 865, "total_steps": 6870, "loss": 0.37, "lr": 4.989896657877972e-05, "epoch": 0.3777292576419214, "percentage": 12.59, "elapsed_time": "7:01:00", "remaining_time": "2 days, 0:42:42"} +{"current_steps": 870, "total_steps": 6870, "loss": 0.3742, "lr": 4.98931819786633e-05, "epoch": 0.3799126637554585, "percentage": 12.66, "elapsed_time": "7:03:26", "remaining_time": "2 days, 0:40:17"} +{"current_steps": 875, "total_steps": 6870, "loss": 0.3658, "lr": 4.988723671348654e-05, "epoch": 0.38209606986899564, "percentage": 12.74, "elapsed_time": "7:05:52", "remaining_time": "2 days, 0:37:52"} +{"current_steps": 880, "total_steps": 6870, "loss": 0.3914, "lr": 4.9881130821621255e-05, "epoch": 0.38427947598253276, "percentage": 12.81, "elapsed_time": "7:08:18", "remaining_time": "2 days, 0:35:26"} +{"current_steps": 885, "total_steps": 6870, "loss": 0.3754, "lr": 4.987486434247596e-05, "epoch": 0.3864628820960699, "percentage": 12.88, "elapsed_time": "7:10:44", "remaining_time": "2 days, 0:33:01"} +{"current_steps": 890, "total_steps": 6870, "loss": 0.3724, "lr": 4.9868437316495636e-05, "epoch": 0.388646288209607, "percentage": 12.95, "elapsed_time": "7:13:10", "remaining_time": "2 days, 0:30:35"} +{"current_steps": 895, "total_steps": 6870, "loss": 0.3673, "lr": 4.9861849785161465e-05, "epoch": 0.39082969432314413, "percentage": 13.03, "elapsed_time": "7:15:37", "remaining_time": "2 days, 0:28:10"} +{"current_steps": 900, "total_steps": 6870, "loss": 0.3727, "lr": 4.9855101790990556e-05, "epoch": 0.3930131004366812, "percentage": 13.1, "elapsed_time": "7:18:03", "remaining_time": "2 days, 0:25:45"} +{"current_steps": 905, "total_steps": 6870, "loss": 0.3841, "lr": 4.9848193377535665e-05, "epoch": 0.3951965065502183, "percentage": 13.17, "elapsed_time": "7:20:29", "remaining_time": "2 days, 0:23:18"} +{"current_steps": 910, "total_steps": 6870, "loss": 0.3723, "lr": 4.9841124589384944e-05, "epoch": 0.39737991266375544, "percentage": 13.25, "elapsed_time": "7:22:54", "remaining_time": "2 days, 0:20:50"} +{"current_steps": 915, "total_steps": 6870, "loss": 0.3707, "lr": 4.983389547216161e-05, "epoch": 0.39956331877729256, "percentage": 13.32, "elapsed_time": "7:25:20", "remaining_time": "2 days, 0:18:24"} +{"current_steps": 920, "total_steps": 6870, "loss": 0.3545, "lr": 4.9826506072523684e-05, "epoch": 0.4017467248908297, "percentage": 13.39, "elapsed_time": "7:27:46", "remaining_time": "2 days, 0:15:57"} +{"current_steps": 925, "total_steps": 6870, "loss": 0.3718, "lr": 4.981895643816367e-05, "epoch": 0.4039301310043668, "percentage": 13.46, "elapsed_time": "7:30:12", "remaining_time": "2 days, 0:13:31"} +{"current_steps": 930, "total_steps": 6870, "loss": 0.3684, "lr": 4.9811246617808275e-05, "epoch": 0.40611353711790393, "percentage": 13.54, "elapsed_time": "7:32:38", "remaining_time": "2 days, 0:11:06"} +{"current_steps": 935, "total_steps": 6870, "loss": 0.3789, "lr": 4.980337666121806e-05, "epoch": 0.40829694323144106, "percentage": 13.61, "elapsed_time": "7:35:05", "remaining_time": "2 days, 0:08:41"} +{"current_steps": 940, "total_steps": 6870, "loss": 0.3542, "lr": 4.979534661918712e-05, "epoch": 0.4104803493449782, "percentage": 13.68, "elapsed_time": "7:37:31", "remaining_time": "2 days, 0:06:16"} +{"current_steps": 945, "total_steps": 6870, "loss": 0.3595, "lr": 4.978715654354281e-05, "epoch": 0.4126637554585153, "percentage": 13.76, "elapsed_time": "7:39:57", "remaining_time": "2 days, 0:03:50"} +{"current_steps": 950, "total_steps": 6870, "loss": 0.3666, "lr": 4.9778806487145326e-05, "epoch": 0.4148471615720524, "percentage": 13.83, "elapsed_time": "7:42:23", "remaining_time": "2 days, 0:01:24"} +{"current_steps": 955, "total_steps": 6870, "loss": 0.3654, "lr": 4.977029650388744e-05, "epoch": 0.4170305676855895, "percentage": 13.9, "elapsed_time": "7:44:49", "remaining_time": "1 day, 23:58:59"} +{"current_steps": 960, "total_steps": 6870, "loss": 0.3542, "lr": 4.976162664869411e-05, "epoch": 0.4192139737991266, "percentage": 13.97, "elapsed_time": "7:47:15", "remaining_time": "1 day, 23:56:34"} +{"current_steps": 965, "total_steps": 6870, "loss": 0.3594, "lr": 4.975279697752213e-05, "epoch": 0.42139737991266374, "percentage": 14.05, "elapsed_time": "7:49:41", "remaining_time": "1 day, 23:54:09"} +{"current_steps": 970, "total_steps": 6870, "loss": 0.3669, "lr": 4.974380754735979e-05, "epoch": 0.42358078602620086, "percentage": 14.12, "elapsed_time": "7:52:07", "remaining_time": "1 day, 23:51:44"} +{"current_steps": 975, "total_steps": 6870, "loss": 0.3616, "lr": 4.9734658416226476e-05, "epoch": 0.425764192139738, "percentage": 14.19, "elapsed_time": "7:54:34", "remaining_time": "1 day, 23:49:18"} +{"current_steps": 980, "total_steps": 6870, "loss": 0.369, "lr": 4.972534964317233e-05, "epoch": 0.4279475982532751, "percentage": 14.26, "elapsed_time": "7:57:00", "remaining_time": "1 day, 23:46:53"} +{"current_steps": 985, "total_steps": 6870, "loss": 0.3594, "lr": 4.9715881288277835e-05, "epoch": 0.43013100436681223, "percentage": 14.34, "elapsed_time": "7:59:26", "remaining_time": "1 day, 23:44:28"} +{"current_steps": 990, "total_steps": 6870, "loss": 0.3549, "lr": 4.9706253412653446e-05, "epoch": 0.43231441048034935, "percentage": 14.41, "elapsed_time": "8:01:52", "remaining_time": "1 day, 23:42:02"} +{"current_steps": 995, "total_steps": 6870, "loss": 0.3671, "lr": 4.9696466078439216e-05, "epoch": 0.4344978165938865, "percentage": 14.48, "elapsed_time": "8:04:18", "remaining_time": "1 day, 23:39:37"} +{"current_steps": 1000, "total_steps": 6870, "loss": 0.365, "lr": 4.968651934880434e-05, "epoch": 0.4366812227074236, "percentage": 14.56, "elapsed_time": "8:06:44", "remaining_time": "1 day, 23:37:11"} +{"current_steps": 1005, "total_steps": 6870, "loss": 0.3598, "lr": 4.96764132879468e-05, "epoch": 0.4388646288209607, "percentage": 14.63, "elapsed_time": "8:09:10", "remaining_time": "1 day, 23:34:46"} +{"current_steps": 1010, "total_steps": 6870, "loss": 0.3659, "lr": 4.9666147961092924e-05, "epoch": 0.4410480349344978, "percentage": 14.7, "elapsed_time": "8:11:36", "remaining_time": "1 day, 23:32:20"} +{"current_steps": 1015, "total_steps": 6870, "loss": 0.3719, "lr": 4.9655723434496984e-05, "epoch": 0.4432314410480349, "percentage": 14.77, "elapsed_time": "8:14:03", "remaining_time": "1 day, 23:29:55"} +{"current_steps": 1020, "total_steps": 6870, "loss": 0.3652, "lr": 4.9645139775440716e-05, "epoch": 0.44541484716157204, "percentage": 14.85, "elapsed_time": "8:16:29", "remaining_time": "1 day, 23:27:30"} +{"current_steps": 1025, "total_steps": 6870, "loss": 0.3691, "lr": 4.9634397052232983e-05, "epoch": 0.44759825327510916, "percentage": 14.92, "elapsed_time": "8:18:55", "remaining_time": "1 day, 23:25:04"} +{"current_steps": 1030, "total_steps": 6870, "loss": 0.3624, "lr": 4.9623495334209226e-05, "epoch": 0.4497816593886463, "percentage": 14.99, "elapsed_time": "8:21:21", "remaining_time": "1 day, 23:22:38"} +{"current_steps": 1035, "total_steps": 6870, "loss": 0.3676, "lr": 4.96124346917311e-05, "epoch": 0.4519650655021834, "percentage": 15.07, "elapsed_time": "8:23:47", "remaining_time": "1 day, 23:20:13"} +{"current_steps": 1040, "total_steps": 6870, "loss": 0.332, "lr": 4.960121519618597e-05, "epoch": 0.45414847161572053, "percentage": 15.14, "elapsed_time": "8:26:13", "remaining_time": "1 day, 23:17:47"} +{"current_steps": 1045, "total_steps": 6870, "loss": 0.3512, "lr": 4.9589836919986475e-05, "epoch": 0.45633187772925765, "percentage": 15.21, "elapsed_time": "8:28:39", "remaining_time": "1 day, 23:15:21"} +{"current_steps": 1050, "total_steps": 6870, "loss": 0.3501, "lr": 4.957829993657005e-05, "epoch": 0.4585152838427948, "percentage": 15.28, "elapsed_time": "8:31:05", "remaining_time": "1 day, 23:12:54"} +{"current_steps": 1055, "total_steps": 6870, "loss": 0.3597, "lr": 4.956660432039846e-05, "epoch": 0.4606986899563319, "percentage": 15.36, "elapsed_time": "8:33:31", "remaining_time": "1 day, 23:10:29"} +{"current_steps": 1060, "total_steps": 6870, "loss": 0.3676, "lr": 4.955475014695729e-05, "epoch": 0.462882096069869, "percentage": 15.43, "elapsed_time": "8:35:57", "remaining_time": "1 day, 23:08:04"} +{"current_steps": 1065, "total_steps": 6870, "loss": 0.3727, "lr": 4.954273749275552e-05, "epoch": 0.4650655021834061, "percentage": 15.5, "elapsed_time": "8:38:23", "remaining_time": "1 day, 23:05:38"} +{"current_steps": 1070, "total_steps": 6870, "loss": 0.3488, "lr": 4.9530566435324963e-05, "epoch": 0.4672489082969432, "percentage": 15.57, "elapsed_time": "8:40:50", "remaining_time": "1 day, 23:03:13"} +{"current_steps": 1075, "total_steps": 6870, "loss": 0.3709, "lr": 4.9518237053219816e-05, "epoch": 0.46943231441048033, "percentage": 15.65, "elapsed_time": "8:43:16", "remaining_time": "1 day, 23:00:47"} +{"current_steps": 1080, "total_steps": 6870, "loss": 0.3475, "lr": 4.950574942601611e-05, "epoch": 0.47161572052401746, "percentage": 15.72, "elapsed_time": "8:45:42", "remaining_time": "1 day, 22:58:22"} +{"current_steps": 1085, "total_steps": 6870, "loss": 0.3527, "lr": 4.9493103634311245e-05, "epoch": 0.4737991266375546, "percentage": 15.79, "elapsed_time": "8:48:08", "remaining_time": "1 day, 22:55:57"} +{"current_steps": 1090, "total_steps": 6870, "loss": 0.34, "lr": 4.948029975972341e-05, "epoch": 0.4759825327510917, "percentage": 15.87, "elapsed_time": "8:50:34", "remaining_time": "1 day, 22:53:31"} +{"current_steps": 1095, "total_steps": 6870, "loss": 0.3578, "lr": 4.9467337884891127e-05, "epoch": 0.4781659388646288, "percentage": 15.94, "elapsed_time": "8:53:00", "remaining_time": "1 day, 22:51:06"} +{"current_steps": 1100, "total_steps": 6870, "loss": 0.3437, "lr": 4.945421809347266e-05, "epoch": 0.48034934497816595, "percentage": 16.01, "elapsed_time": "8:55:27", "remaining_time": "1 day, 22:48:40"} +{"current_steps": 1105, "total_steps": 6870, "loss": 0.3582, "lr": 4.944094047014547e-05, "epoch": 0.48253275109170307, "percentage": 16.08, "elapsed_time": "8:57:53", "remaining_time": "1 day, 22:46:14"} +{"current_steps": 1110, "total_steps": 6870, "loss": 0.3369, "lr": 4.942750510060574e-05, "epoch": 0.4847161572052402, "percentage": 16.16, "elapsed_time": "9:00:19", "remaining_time": "1 day, 22:43:48"} +{"current_steps": 1115, "total_steps": 6870, "loss": 0.3683, "lr": 4.9413912071567756e-05, "epoch": 0.4868995633187773, "percentage": 16.23, "elapsed_time": "9:02:45", "remaining_time": "1 day, 22:41:22"} +{"current_steps": 1120, "total_steps": 6870, "loss": 0.3425, "lr": 4.940016147076336e-05, "epoch": 0.4890829694323144, "percentage": 16.3, "elapsed_time": "9:05:11", "remaining_time": "1 day, 22:38:56"} +{"current_steps": 1125, "total_steps": 6870, "loss": 0.3609, "lr": 4.938625338694139e-05, "epoch": 0.4912663755458515, "percentage": 16.38, "elapsed_time": "9:07:37", "remaining_time": "1 day, 22:36:30"} +{"current_steps": 1130, "total_steps": 6870, "loss": 0.3532, "lr": 4.937218790986712e-05, "epoch": 0.49344978165938863, "percentage": 16.45, "elapsed_time": "9:10:02", "remaining_time": "1 day, 22:34:02"} +{"current_steps": 1135, "total_steps": 6870, "loss": 0.3434, "lr": 4.935796513032165e-05, "epoch": 0.49563318777292575, "percentage": 16.52, "elapsed_time": "9:12:28", "remaining_time": "1 day, 22:31:37"} +{"current_steps": 1140, "total_steps": 6870, "loss": 0.3569, "lr": 4.934358514010137e-05, "epoch": 0.4978165938864629, "percentage": 16.59, "elapsed_time": "9:14:55", "remaining_time": "1 day, 22:29:11"} +{"current_steps": 1145, "total_steps": 6870, "loss": 0.3545, "lr": 4.93290480320173e-05, "epoch": 0.5, "percentage": 16.67, "elapsed_time": "9:17:21", "remaining_time": "1 day, 22:26:45"} +{"current_steps": 1150, "total_steps": 6870, "loss": 0.3482, "lr": 4.931435389989455e-05, "epoch": 0.5021834061135371, "percentage": 16.74, "elapsed_time": "9:19:47", "remaining_time": "1 day, 22:24:20"} +{"current_steps": 1155, "total_steps": 6870, "loss": 0.3704, "lr": 4.929950283857167e-05, "epoch": 0.5043668122270742, "percentage": 16.81, "elapsed_time": "9:22:13", "remaining_time": "1 day, 22:21:55"} +{"current_steps": 1160, "total_steps": 6870, "loss": 0.3646, "lr": 4.9284494943900085e-05, "epoch": 0.5065502183406113, "percentage": 16.89, "elapsed_time": "9:24:39", "remaining_time": "1 day, 22:19:29"} +{"current_steps": 1165, "total_steps": 6870, "loss": 0.3626, "lr": 4.926933031274344e-05, "epoch": 0.5087336244541485, "percentage": 16.96, "elapsed_time": "9:27:05", "remaining_time": "1 day, 22:17:04"} +{"current_steps": 1170, "total_steps": 6870, "loss": 0.347, "lr": 4.925400904297698e-05, "epoch": 0.5109170305676856, "percentage": 17.03, "elapsed_time": "9:29:32", "remaining_time": "1 day, 22:14:39"} +{"current_steps": 1175, "total_steps": 6870, "loss": 0.3678, "lr": 4.923853123348693e-05, "epoch": 0.5131004366812227, "percentage": 17.1, "elapsed_time": "9:31:58", "remaining_time": "1 day, 22:12:13"} +{"current_steps": 1180, "total_steps": 6870, "loss": 0.3568, "lr": 4.922289698416983e-05, "epoch": 0.5152838427947598, "percentage": 17.18, "elapsed_time": "9:34:24", "remaining_time": "1 day, 22:09:48"} +{"current_steps": 1185, "total_steps": 6870, "loss": 0.3555, "lr": 4.920710639593196e-05, "epoch": 0.517467248908297, "percentage": 17.25, "elapsed_time": "9:36:50", "remaining_time": "1 day, 22:07:22"} +{"current_steps": 1190, "total_steps": 6870, "loss": 0.3529, "lr": 4.919115957068859e-05, "epoch": 0.519650655021834, "percentage": 17.32, "elapsed_time": "9:39:16", "remaining_time": "1 day, 22:04:57"} +{"current_steps": 1195, "total_steps": 6870, "loss": 0.3491, "lr": 4.917505661136339e-05, "epoch": 0.5218340611353712, "percentage": 17.39, "elapsed_time": "9:41:42", "remaining_time": "1 day, 22:02:31"} +{"current_steps": 1200, "total_steps": 6870, "loss": 0.3615, "lr": 4.915879762188774e-05, "epoch": 0.5240174672489083, "percentage": 17.47, "elapsed_time": "9:44:09", "remaining_time": "1 day, 22:00:06"} +{"current_steps": 1205, "total_steps": 6870, "loss": 0.3553, "lr": 4.9142382707200096e-05, "epoch": 0.5262008733624454, "percentage": 17.54, "elapsed_time": "9:46:35", "remaining_time": "1 day, 21:57:40"} +{"current_steps": 1210, "total_steps": 6870, "loss": 0.3633, "lr": 4.912581197324524e-05, "epoch": 0.5283842794759825, "percentage": 17.61, "elapsed_time": "9:49:01", "remaining_time": "1 day, 21:55:15"} +{"current_steps": 1215, "total_steps": 6870, "loss": 0.3476, "lr": 4.9109085526973674e-05, "epoch": 0.5305676855895196, "percentage": 17.69, "elapsed_time": "9:51:27", "remaining_time": "1 day, 21:52:49"} +{"current_steps": 1220, "total_steps": 6870, "loss": 0.359, "lr": 4.909220347634087e-05, "epoch": 0.5327510917030568, "percentage": 17.76, "elapsed_time": "9:53:53", "remaining_time": "1 day, 21:50:24"} +{"current_steps": 1225, "total_steps": 6870, "loss": 0.3515, "lr": 4.907516593030662e-05, "epoch": 0.5349344978165939, "percentage": 17.83, "elapsed_time": "9:56:19", "remaining_time": "1 day, 21:47:59"} +{"current_steps": 1230, "total_steps": 6870, "loss": 0.3629, "lr": 4.905797299883429e-05, "epoch": 0.537117903930131, "percentage": 17.9, "elapsed_time": "9:58:46", "remaining_time": "1 day, 21:45:33"} +{"current_steps": 1235, "total_steps": 6870, "loss": 0.3484, "lr": 4.904062479289015e-05, "epoch": 0.5393013100436681, "percentage": 17.98, "elapsed_time": "10:01:12", "remaining_time": "1 day, 21:43:07"} +{"current_steps": 1240, "total_steps": 6870, "loss": 0.3526, "lr": 4.902312142444264e-05, "epoch": 0.5414847161572053, "percentage": 18.05, "elapsed_time": "10:03:38", "remaining_time": "1 day, 21:40:41"} +{"current_steps": 1245, "total_steps": 6870, "loss": 0.3366, "lr": 4.900546300646162e-05, "epoch": 0.5436681222707423, "percentage": 18.12, "elapsed_time": "10:06:04", "remaining_time": "1 day, 21:38:15"} +{"current_steps": 1250, "total_steps": 6870, "loss": 0.3543, "lr": 4.89876496529177e-05, "epoch": 0.5458515283842795, "percentage": 18.2, "elapsed_time": "10:08:30", "remaining_time": "1 day, 21:35:49"} +{"current_steps": 1255, "total_steps": 6870, "loss": 0.3565, "lr": 4.896968147878145e-05, "epoch": 0.5480349344978166, "percentage": 18.27, "elapsed_time": "10:10:56", "remaining_time": "1 day, 21:33:24"} +{"current_steps": 1260, "total_steps": 6870, "loss": 0.3377, "lr": 4.895155860002271e-05, "epoch": 0.5502183406113537, "percentage": 18.34, "elapsed_time": "10:13:22", "remaining_time": "1 day, 21:30:58"} +{"current_steps": 1265, "total_steps": 6870, "loss": 0.3535, "lr": 4.893328113360975e-05, "epoch": 0.5524017467248908, "percentage": 18.41, "elapsed_time": "10:15:48", "remaining_time": "1 day, 21:28:33"} +{"current_steps": 1270, "total_steps": 6870, "loss": 0.344, "lr": 4.891484919750865e-05, "epoch": 0.5545851528384279, "percentage": 18.49, "elapsed_time": "10:18:14", "remaining_time": "1 day, 21:26:07"} +{"current_steps": 1275, "total_steps": 6870, "loss": 0.339, "lr": 4.889626291068241e-05, "epoch": 0.5567685589519651, "percentage": 18.56, "elapsed_time": "10:20:40", "remaining_time": "1 day, 21:23:41"} +{"current_steps": 1280, "total_steps": 6870, "loss": 0.3391, "lr": 4.887752239309027e-05, "epoch": 0.5589519650655022, "percentage": 18.63, "elapsed_time": "10:23:06", "remaining_time": "1 day, 21:21:15"} +{"current_steps": 1285, "total_steps": 6870, "loss": 0.3576, "lr": 4.885862776568689e-05, "epoch": 0.5611353711790393, "percentage": 18.7, "elapsed_time": "10:25:33", "remaining_time": "1 day, 21:18:49"} +{"current_steps": 1290, "total_steps": 6870, "loss": 0.3496, "lr": 4.883957915042158e-05, "epoch": 0.5633187772925764, "percentage": 18.78, "elapsed_time": "10:27:59", "remaining_time": "1 day, 21:16:24"} +{"current_steps": 1295, "total_steps": 6870, "loss": 0.3361, "lr": 4.882037667023752e-05, "epoch": 0.5655021834061136, "percentage": 18.85, "elapsed_time": "10:30:25", "remaining_time": "1 day, 21:13:58"} +{"current_steps": 1300, "total_steps": 6870, "loss": 0.3677, "lr": 4.880102044907096e-05, "epoch": 0.5676855895196506, "percentage": 18.92, "elapsed_time": "10:32:51", "remaining_time": "1 day, 21:11:33"} +{"current_steps": 1305, "total_steps": 6870, "loss": 0.3294, "lr": 4.878151061185042e-05, "epoch": 0.5698689956331878, "percentage": 19.0, "elapsed_time": "10:35:17", "remaining_time": "1 day, 21:09:07"} +{"current_steps": 1310, "total_steps": 6870, "loss": 0.3488, "lr": 4.876184728449589e-05, "epoch": 0.5720524017467249, "percentage": 19.07, "elapsed_time": "10:37:43", "remaining_time": "1 day, 21:06:41"} +{"current_steps": 1315, "total_steps": 6870, "loss": 0.3364, "lr": 4.874203059391802e-05, "epoch": 0.574235807860262, "percentage": 19.14, "elapsed_time": "10:40:09", "remaining_time": "1 day, 21:04:16"} +{"current_steps": 1320, "total_steps": 6870, "loss": 0.3562, "lr": 4.8722060668017267e-05, "epoch": 0.5764192139737991, "percentage": 19.21, "elapsed_time": "10:42:36", "remaining_time": "1 day, 21:01:51"} +{"current_steps": 1325, "total_steps": 6870, "loss": 0.3619, "lr": 4.870193763568312e-05, "epoch": 0.5786026200873362, "percentage": 19.29, "elapsed_time": "10:45:02", "remaining_time": "1 day, 20:59:25"} +{"current_steps": 1330, "total_steps": 6870, "loss": 0.3468, "lr": 4.8681661626793246e-05, "epoch": 0.5807860262008734, "percentage": 19.36, "elapsed_time": "10:47:28", "remaining_time": "1 day, 20:56:59"} +{"current_steps": 1335, "total_steps": 6870, "loss": 0.347, "lr": 4.866123277221264e-05, "epoch": 0.5829694323144105, "percentage": 19.43, "elapsed_time": "10:49:54", "remaining_time": "1 day, 20:54:33"} +{"current_steps": 1340, "total_steps": 6870, "loss": 0.3573, "lr": 4.864065120379278e-05, "epoch": 0.5851528384279476, "percentage": 19.51, "elapsed_time": "10:52:20", "remaining_time": "1 day, 20:52:07"} +{"current_steps": 1345, "total_steps": 6870, "loss": 0.3598, "lr": 4.861991705437081e-05, "epoch": 0.5873362445414847, "percentage": 19.58, "elapsed_time": "10:54:46", "remaining_time": "1 day, 20:49:42"} +{"current_steps": 1350, "total_steps": 6870, "loss": 0.3524, "lr": 4.859903045776866e-05, "epoch": 0.5895196506550219, "percentage": 19.65, "elapsed_time": "10:57:13", "remaining_time": "1 day, 20:47:17"} +{"current_steps": 1355, "total_steps": 6870, "loss": 0.3605, "lr": 4.857799154879215e-05, "epoch": 0.5917030567685589, "percentage": 19.72, "elapsed_time": "10:59:39", "remaining_time": "1 day, 20:44:52"} +{"current_steps": 1360, "total_steps": 6870, "loss": 0.3437, "lr": 4.8556800463230165e-05, "epoch": 0.5938864628820961, "percentage": 19.8, "elapsed_time": "11:02:05", "remaining_time": "1 day, 20:42:26"} +{"current_steps": 1365, "total_steps": 6870, "loss": 0.3285, "lr": 4.8535457337853794e-05, "epoch": 0.5960698689956332, "percentage": 19.87, "elapsed_time": "11:04:31", "remaining_time": "1 day, 20:40:01"} +{"current_steps": 1370, "total_steps": 6870, "loss": 0.3463, "lr": 4.8513962310415396e-05, "epoch": 0.5982532751091703, "percentage": 19.94, "elapsed_time": "11:06:57", "remaining_time": "1 day, 20:37:35"} +{"current_steps": 1375, "total_steps": 6870, "loss": 0.3535, "lr": 4.849231551964771e-05, "epoch": 0.6004366812227074, "percentage": 20.01, "elapsed_time": "11:09:23", "remaining_time": "1 day, 20:35:09"} +{"current_steps": 1380, "total_steps": 6870, "loss": 0.3437, "lr": 4.847051710526302e-05, "epoch": 0.6026200873362445, "percentage": 20.09, "elapsed_time": "11:11:50", "remaining_time": "1 day, 20:32:44"} +{"current_steps": 1385, "total_steps": 6870, "loss": 0.3549, "lr": 4.8448567207952205e-05, "epoch": 0.6048034934497817, "percentage": 20.16, "elapsed_time": "11:14:16", "remaining_time": "1 day, 20:30:19"} +{"current_steps": 1390, "total_steps": 6870, "loss": 0.3243, "lr": 4.8426465969383826e-05, "epoch": 0.6069868995633187, "percentage": 20.23, "elapsed_time": "11:16:42", "remaining_time": "1 day, 20:27:53"} +{"current_steps": 1395, "total_steps": 6870, "loss": 0.3251, "lr": 4.840421353220326e-05, "epoch": 0.6091703056768559, "percentage": 20.31, "elapsed_time": "11:19:08", "remaining_time": "1 day, 20:25:28"} +{"current_steps": 1400, "total_steps": 6870, "loss": 0.334, "lr": 4.83818100400317e-05, "epoch": 0.611353711790393, "percentage": 20.38, "elapsed_time": "11:21:35", "remaining_time": "1 day, 20:23:03"} +{"current_steps": 1405, "total_steps": 6870, "loss": 0.3551, "lr": 4.835925563746532e-05, "epoch": 0.6135371179039302, "percentage": 20.45, "elapsed_time": "11:24:01", "remaining_time": "1 day, 20:20:37"} +{"current_steps": 1410, "total_steps": 6870, "loss": 0.3504, "lr": 4.833655047007427e-05, "epoch": 0.6157205240174672, "percentage": 20.52, "elapsed_time": "11:26:27", "remaining_time": "1 day, 20:18:11"} +{"current_steps": 1415, "total_steps": 6870, "loss": 0.3358, "lr": 4.831369468440177e-05, "epoch": 0.6179039301310044, "percentage": 20.6, "elapsed_time": "11:28:53", "remaining_time": "1 day, 20:15:45"} +{"current_steps": 1420, "total_steps": 6870, "loss": 0.3315, "lr": 4.8290688427963175e-05, "epoch": 0.6200873362445415, "percentage": 20.67, "elapsed_time": "11:31:19", "remaining_time": "1 day, 20:13:20"} +{"current_steps": 1425, "total_steps": 6870, "loss": 0.3403, "lr": 4.8267531849244974e-05, "epoch": 0.6222707423580786, "percentage": 20.74, "elapsed_time": "11:33:45", "remaining_time": "1 day, 20:10:54"} +{"current_steps": 1430, "total_steps": 6870, "loss": 0.351, "lr": 4.8244225097703885e-05, "epoch": 0.6244541484716157, "percentage": 20.82, "elapsed_time": "11:36:12", "remaining_time": "1 day, 20:08:29"} +{"current_steps": 1435, "total_steps": 6870, "loss": 0.345, "lr": 4.8220768323765866e-05, "epoch": 0.6266375545851528, "percentage": 20.89, "elapsed_time": "11:38:38", "remaining_time": "1 day, 20:06:03"} +{"current_steps": 1440, "total_steps": 6870, "loss": 0.343, "lr": 4.819716167882513e-05, "epoch": 0.62882096069869, "percentage": 20.96, "elapsed_time": "11:41:04", "remaining_time": "1 day, 20:03:37"} +{"current_steps": 1445, "total_steps": 6870, "loss": 0.3381, "lr": 4.8173405315243216e-05, "epoch": 0.631004366812227, "percentage": 21.03, "elapsed_time": "11:43:30", "remaining_time": "1 day, 20:01:11"} +{"current_steps": 1450, "total_steps": 6870, "loss": 0.3334, "lr": 4.814949938634793e-05, "epoch": 0.6331877729257642, "percentage": 21.11, "elapsed_time": "11:45:56", "remaining_time": "1 day, 19:58:45"} +{"current_steps": 1455, "total_steps": 6870, "loss": 0.3387, "lr": 4.8125444046432434e-05, "epoch": 0.6353711790393013, "percentage": 21.18, "elapsed_time": "11:48:22", "remaining_time": "1 day, 19:56:19"} +{"current_steps": 1460, "total_steps": 6870, "loss": 0.3485, "lr": 4.810123945075421e-05, "epoch": 0.6375545851528385, "percentage": 21.25, "elapsed_time": "11:50:48", "remaining_time": "1 day, 19:53:53"} +{"current_steps": 1465, "total_steps": 6870, "loss": 0.3381, "lr": 4.807688575553406e-05, "epoch": 0.6397379912663755, "percentage": 21.32, "elapsed_time": "11:53:14", "remaining_time": "1 day, 19:51:28"} +{"current_steps": 1470, "total_steps": 6870, "loss": 0.3316, "lr": 4.80523831179551e-05, "epoch": 0.6419213973799127, "percentage": 21.4, "elapsed_time": "11:55:40", "remaining_time": "1 day, 19:49:02"} +{"current_steps": 1475, "total_steps": 6870, "loss": 0.3378, "lr": 4.8027731696161735e-05, "epoch": 0.6441048034934498, "percentage": 21.47, "elapsed_time": "11:58:07", "remaining_time": "1 day, 19:46:36"} +{"current_steps": 1480, "total_steps": 6870, "loss": 0.3343, "lr": 4.8002931649258674e-05, "epoch": 0.6462882096069869, "percentage": 21.54, "elapsed_time": "12:00:33", "remaining_time": "1 day, 19:44:10"} +{"current_steps": 1485, "total_steps": 6870, "loss": 0.3337, "lr": 4.7977983137309855e-05, "epoch": 0.648471615720524, "percentage": 21.62, "elapsed_time": "12:02:59", "remaining_time": "1 day, 19:41:45"} +{"current_steps": 1490, "total_steps": 6870, "loss": 0.3328, "lr": 4.795288632133747e-05, "epoch": 0.6506550218340611, "percentage": 21.69, "elapsed_time": "12:05:25", "remaining_time": "1 day, 19:39:19"} +{"current_steps": 1495, "total_steps": 6870, "loss": 0.3558, "lr": 4.792764136332084e-05, "epoch": 0.6528384279475983, "percentage": 21.76, "elapsed_time": "12:07:51", "remaining_time": "1 day, 19:36:53"} +{"current_steps": 1500, "total_steps": 6870, "loss": 0.3305, "lr": 4.790224842619548e-05, "epoch": 0.6550218340611353, "percentage": 21.83, "elapsed_time": "12:10:17", "remaining_time": "1 day, 19:34:27"} +{"current_steps": 1505, "total_steps": 6870, "loss": 0.3399, "lr": 4.787670767385194e-05, "epoch": 0.6572052401746725, "percentage": 21.91, "elapsed_time": "12:12:43", "remaining_time": "1 day, 19:32:01"} +{"current_steps": 1510, "total_steps": 6870, "loss": 0.3423, "lr": 4.785101927113482e-05, "epoch": 0.6593886462882096, "percentage": 21.98, "elapsed_time": "12:15:09", "remaining_time": "1 day, 19:29:35"} +{"current_steps": 1515, "total_steps": 6870, "loss": 0.3474, "lr": 4.782518338384168e-05, "epoch": 0.6615720524017468, "percentage": 22.05, "elapsed_time": "12:17:36", "remaining_time": "1 day, 19:27:10"} +{"current_steps": 1520, "total_steps": 6870, "loss": 0.3426, "lr": 4.779920017872197e-05, "epoch": 0.6637554585152838, "percentage": 22.13, "elapsed_time": "12:20:02", "remaining_time": "1 day, 19:24:44"} +{"current_steps": 1525, "total_steps": 6870, "loss": 0.3317, "lr": 4.7773069823475945e-05, "epoch": 0.665938864628821, "percentage": 22.2, "elapsed_time": "12:22:28", "remaining_time": "1 day, 19:22:18"} +{"current_steps": 1530, "total_steps": 6870, "loss": 0.3468, "lr": 4.774679248675362e-05, "epoch": 0.6681222707423581, "percentage": 22.27, "elapsed_time": "12:24:54", "remaining_time": "1 day, 19:19:52"} +{"current_steps": 1535, "total_steps": 6870, "loss": 0.3305, "lr": 4.772036833815363e-05, "epoch": 0.6703056768558951, "percentage": 22.34, "elapsed_time": "12:27:20", "remaining_time": "1 day, 19:17:26"} +{"current_steps": 1540, "total_steps": 6870, "loss": 0.3226, "lr": 4.7693797548222166e-05, "epoch": 0.6724890829694323, "percentage": 22.42, "elapsed_time": "12:29:46", "remaining_time": "1 day, 19:15:00"} +{"current_steps": 1545, "total_steps": 6870, "loss": 0.3325, "lr": 4.766708028845188e-05, "epoch": 0.6746724890829694, "percentage": 22.49, "elapsed_time": "12:32:12", "remaining_time": "1 day, 19:12:35"} +{"current_steps": 1550, "total_steps": 6870, "loss": 0.3482, "lr": 4.764021673128075e-05, "epoch": 0.6768558951965066, "percentage": 22.56, "elapsed_time": "12:34:39", "remaining_time": "1 day, 19:10:09"} +{"current_steps": 1555, "total_steps": 6870, "loss": 0.3457, "lr": 4.761320705009099e-05, "epoch": 0.6790393013100436, "percentage": 22.63, "elapsed_time": "12:37:05", "remaining_time": "1 day, 19:07:43"} +{"current_steps": 1560, "total_steps": 6870, "loss": 0.3512, "lr": 4.758605141920792e-05, "epoch": 0.6812227074235808, "percentage": 22.71, "elapsed_time": "12:39:31", "remaining_time": "1 day, 19:05:17"} +{"current_steps": 1565, "total_steps": 6870, "loss": 0.3434, "lr": 4.755875001389887e-05, "epoch": 0.6834061135371179, "percentage": 22.78, "elapsed_time": "12:41:57", "remaining_time": "1 day, 19:02:50"} +{"current_steps": 1570, "total_steps": 6870, "loss": 0.3337, "lr": 4.753130301037199e-05, "epoch": 0.6855895196506551, "percentage": 22.85, "elapsed_time": "12:44:23", "remaining_time": "1 day, 19:00:24"} +{"current_steps": 1575, "total_steps": 6870, "loss": 0.3387, "lr": 4.7503710585775174e-05, "epoch": 0.6877729257641921, "percentage": 22.93, "elapsed_time": "12:46:49", "remaining_time": "1 day, 18:57:59"} +{"current_steps": 1580, "total_steps": 6870, "loss": 0.3439, "lr": 4.747597291819487e-05, "epoch": 0.6899563318777293, "percentage": 23.0, "elapsed_time": "12:49:15", "remaining_time": "1 day, 18:55:33"} +{"current_steps": 1585, "total_steps": 6870, "loss": 0.3373, "lr": 4.744809018665495e-05, "epoch": 0.6921397379912664, "percentage": 23.07, "elapsed_time": "12:51:41", "remaining_time": "1 day, 18:53:07"} +{"current_steps": 1590, "total_steps": 6870, "loss": 0.3358, "lr": 4.742006257111558e-05, "epoch": 0.6943231441048034, "percentage": 23.14, "elapsed_time": "12:54:07", "remaining_time": "1 day, 18:50:41"} +{"current_steps": 1595, "total_steps": 6870, "loss": 0.3624, "lr": 4.7391890252472016e-05, "epoch": 0.6965065502183406, "percentage": 23.22, "elapsed_time": "12:56:33", "remaining_time": "1 day, 18:48:15"} +{"current_steps": 1600, "total_steps": 6870, "loss": 0.3427, "lr": 4.736357341255344e-05, "epoch": 0.6986899563318777, "percentage": 23.29, "elapsed_time": "12:58:59", "remaining_time": "1 day, 18:45:49"} +{"current_steps": 1605, "total_steps": 6870, "loss": 0.3332, "lr": 4.7335112234121824e-05, "epoch": 0.7008733624454149, "percentage": 23.36, "elapsed_time": "13:01:26", "remaining_time": "1 day, 18:43:23"} +{"current_steps": 1610, "total_steps": 6870, "loss": 0.3406, "lr": 4.730650690087072e-05, "epoch": 0.7030567685589519, "percentage": 23.44, "elapsed_time": "13:03:52", "remaining_time": "1 day, 18:40:57"} +{"current_steps": 1615, "total_steps": 6870, "loss": 0.3431, "lr": 4.727775759742408e-05, "epoch": 0.7052401746724891, "percentage": 23.51, "elapsed_time": "13:06:18", "remaining_time": "1 day, 18:38:31"} +{"current_steps": 1620, "total_steps": 6870, "loss": 0.3436, "lr": 4.724886450933505e-05, "epoch": 0.7074235807860262, "percentage": 23.58, "elapsed_time": "13:08:44", "remaining_time": "1 day, 18:36:06"} +{"current_steps": 1625, "total_steps": 6870, "loss": 0.3523, "lr": 4.721982782308481e-05, "epoch": 0.7096069868995634, "percentage": 23.65, "elapsed_time": "13:11:10", "remaining_time": "1 day, 18:33:40"} +{"current_steps": 1630, "total_steps": 6870, "loss": 0.3462, "lr": 4.719064772608134e-05, "epoch": 0.7117903930131004, "percentage": 23.73, "elapsed_time": "13:13:36", "remaining_time": "1 day, 18:31:14"} +{"current_steps": 1635, "total_steps": 6870, "loss": 0.3286, "lr": 4.7161324406658225e-05, "epoch": 0.7139737991266376, "percentage": 23.8, "elapsed_time": "13:16:02", "remaining_time": "1 day, 18:28:48"} +{"current_steps": 1640, "total_steps": 6870, "loss": 0.3236, "lr": 4.713185805407342e-05, "epoch": 0.7161572052401747, "percentage": 23.87, "elapsed_time": "13:18:28", "remaining_time": "1 day, 18:26:22"} +{"current_steps": 1645, "total_steps": 6870, "loss": 0.3544, "lr": 4.7102248858508044e-05, "epoch": 0.7183406113537117, "percentage": 23.94, "elapsed_time": "13:20:54", "remaining_time": "1 day, 18:23:56"} +{"current_steps": 1650, "total_steps": 6870, "loss": 0.3298, "lr": 4.7072497011065156e-05, "epoch": 0.7205240174672489, "percentage": 24.02, "elapsed_time": "13:23:21", "remaining_time": "1 day, 18:21:30"} +{"current_steps": 1655, "total_steps": 6870, "loss": 0.3264, "lr": 4.704260270376853e-05, "epoch": 0.722707423580786, "percentage": 24.09, "elapsed_time": "13:25:47", "remaining_time": "1 day, 18:19:04"} +{"current_steps": 1660, "total_steps": 6870, "loss": 0.3381, "lr": 4.7012566129561365e-05, "epoch": 0.7248908296943232, "percentage": 24.16, "elapsed_time": "13:28:13", "remaining_time": "1 day, 18:16:38"} +{"current_steps": 1665, "total_steps": 6870, "loss": 0.3322, "lr": 4.6982387482305104e-05, "epoch": 0.7270742358078602, "percentage": 24.24, "elapsed_time": "13:30:39", "remaining_time": "1 day, 18:14:12"} +{"current_steps": 1670, "total_steps": 6870, "loss": 0.3268, "lr": 4.6952066956778155e-05, "epoch": 0.7292576419213974, "percentage": 24.31, "elapsed_time": "13:33:05", "remaining_time": "1 day, 18:11:46"} +{"current_steps": 1675, "total_steps": 6870, "loss": 0.3324, "lr": 4.692160474867463e-05, "epoch": 0.7314410480349345, "percentage": 24.38, "elapsed_time": "13:35:31", "remaining_time": "1 day, 18:09:20"} +{"current_steps": 1680, "total_steps": 6870, "loss": 0.3374, "lr": 4.6891001054603056e-05, "epoch": 0.7336244541484717, "percentage": 24.45, "elapsed_time": "13:37:57", "remaining_time": "1 day, 18:06:55"} +{"current_steps": 1685, "total_steps": 6870, "loss": 0.3453, "lr": 4.6860256072085196e-05, "epoch": 0.7358078602620087, "percentage": 24.53, "elapsed_time": "13:40:23", "remaining_time": "1 day, 18:04:29"} +{"current_steps": 1690, "total_steps": 6870, "loss": 0.3297, "lr": 4.6829369999554664e-05, "epoch": 0.7379912663755459, "percentage": 24.6, "elapsed_time": "13:42:50", "remaining_time": "1 day, 18:02:03"} +{"current_steps": 1695, "total_steps": 6870, "loss": 0.3058, "lr": 4.6798343036355704e-05, "epoch": 0.740174672489083, "percentage": 24.67, "elapsed_time": "13:45:16", "remaining_time": "1 day, 17:59:37"} +{"current_steps": 1700, "total_steps": 6870, "loss": 0.3367, "lr": 4.6767175382741914e-05, "epoch": 0.74235807860262, "percentage": 24.75, "elapsed_time": "13:47:42", "remaining_time": "1 day, 17:57:11"} +{"current_steps": 1705, "total_steps": 6870, "loss": 0.3504, "lr": 4.673586723987491e-05, "epoch": 0.7445414847161572, "percentage": 24.82, "elapsed_time": "13:50:08", "remaining_time": "1 day, 17:54:46"} +{"current_steps": 1710, "total_steps": 6870, "loss": 0.3323, "lr": 4.670441880982307e-05, "epoch": 0.7467248908296943, "percentage": 24.89, "elapsed_time": "13:52:34", "remaining_time": "1 day, 17:52:20"} +{"current_steps": 1715, "total_steps": 6870, "loss": 0.3288, "lr": 4.667283029556019e-05, "epoch": 0.7489082969432315, "percentage": 24.96, "elapsed_time": "13:55:00", "remaining_time": "1 day, 17:49:54"} +{"current_steps": 1720, "total_steps": 6870, "loss": 0.3436, "lr": 4.664110190096421e-05, "epoch": 0.7510917030567685, "percentage": 25.04, "elapsed_time": "13:57:26", "remaining_time": "1 day, 17:47:28"} +{"current_steps": 1725, "total_steps": 6870, "loss": 0.3401, "lr": 4.660923383081588e-05, "epoch": 0.7532751091703057, "percentage": 25.11, "elapsed_time": "13:59:52", "remaining_time": "1 day, 17:45:02"} +{"current_steps": 1730, "total_steps": 6870, "loss": 0.3359, "lr": 4.657722629079744e-05, "epoch": 0.7554585152838428, "percentage": 25.18, "elapsed_time": "14:02:19", "remaining_time": "1 day, 17:42:36"} +{"current_steps": 1735, "total_steps": 6870, "loss": 0.3466, "lr": 4.6545079487491285e-05, "epoch": 0.75764192139738, "percentage": 25.25, "elapsed_time": "14:04:45", "remaining_time": "1 day, 17:40:10"} +{"current_steps": 1740, "total_steps": 6870, "loss": 0.3348, "lr": 4.6512793628378657e-05, "epoch": 0.759825327510917, "percentage": 25.33, "elapsed_time": "14:07:11", "remaining_time": "1 day, 17:37:44"} +{"current_steps": 1745, "total_steps": 6870, "loss": 0.3381, "lr": 4.648036892183828e-05, "epoch": 0.7620087336244541, "percentage": 25.4, "elapsed_time": "14:09:37", "remaining_time": "1 day, 17:35:18"} +{"current_steps": 1750, "total_steps": 6870, "loss": 0.3338, "lr": 4.644780557714502e-05, "epoch": 0.7641921397379913, "percentage": 25.47, "elapsed_time": "14:12:03", "remaining_time": "1 day, 17:32:52"} +{"current_steps": 1755, "total_steps": 6870, "loss": 0.3238, "lr": 4.641510380446854e-05, "epoch": 0.7663755458515283, "percentage": 25.55, "elapsed_time": "14:14:29", "remaining_time": "1 day, 17:30:27"} +{"current_steps": 1760, "total_steps": 6870, "loss": 0.3373, "lr": 4.638226381487195e-05, "epoch": 0.7685589519650655, "percentage": 25.62, "elapsed_time": "14:16:55", "remaining_time": "1 day, 17:28:01"} +{"current_steps": 1765, "total_steps": 6870, "loss": 0.3449, "lr": 4.634928582031043e-05, "epoch": 0.7707423580786026, "percentage": 25.69, "elapsed_time": "14:19:21", "remaining_time": "1 day, 17:25:35"} +{"current_steps": 1770, "total_steps": 6870, "loss": 0.3337, "lr": 4.631617003362988e-05, "epoch": 0.7729257641921398, "percentage": 25.76, "elapsed_time": "14:21:48", "remaining_time": "1 day, 17:23:09"} +{"current_steps": 1775, "total_steps": 6870, "loss": 0.3211, "lr": 4.628291666856551e-05, "epoch": 0.7751091703056768, "percentage": 25.84, "elapsed_time": "14:24:14", "remaining_time": "1 day, 17:20:43"} +{"current_steps": 1780, "total_steps": 6870, "loss": 0.3306, "lr": 4.6249525939740514e-05, "epoch": 0.777292576419214, "percentage": 25.91, "elapsed_time": "14:26:40", "remaining_time": "1 day, 17:18:18"} +{"current_steps": 1785, "total_steps": 6870, "loss": 0.336, "lr": 4.6215998062664636e-05, "epoch": 0.7794759825327511, "percentage": 25.98, "elapsed_time": "14:29:06", "remaining_time": "1 day, 17:15:52"} +{"current_steps": 1790, "total_steps": 6870, "loss": 0.3271, "lr": 4.61823332537328e-05, "epoch": 0.7816593886462883, "percentage": 26.06, "elapsed_time": "14:31:32", "remaining_time": "1 day, 17:13:26"} +{"current_steps": 1795, "total_steps": 6870, "loss": 0.3309, "lr": 4.6148531730223735e-05, "epoch": 0.7838427947598253, "percentage": 26.13, "elapsed_time": "14:33:59", "remaining_time": "1 day, 17:11:00"} +{"current_steps": 1800, "total_steps": 6870, "loss": 0.3273, "lr": 4.611459371029853e-05, "epoch": 0.7860262008733624, "percentage": 26.2, "elapsed_time": "14:36:25", "remaining_time": "1 day, 17:08:34"} +{"current_steps": 1805, "total_steps": 6870, "loss": 0.3256, "lr": 4.6080519412999244e-05, "epoch": 0.7882096069868996, "percentage": 26.27, "elapsed_time": "14:38:51", "remaining_time": "1 day, 17:06:08"} +{"current_steps": 1810, "total_steps": 6870, "loss": 0.3344, "lr": 4.604630905824751e-05, "epoch": 0.7903930131004366, "percentage": 26.35, "elapsed_time": "14:41:17", "remaining_time": "1 day, 17:03:42"} +{"current_steps": 1815, "total_steps": 6870, "loss": 0.3301, "lr": 4.601196286684309e-05, "epoch": 0.7925764192139738, "percentage": 26.42, "elapsed_time": "14:43:43", "remaining_time": "1 day, 17:01:16"} +{"current_steps": 1820, "total_steps": 6870, "loss": 0.3305, "lr": 4.597748106046247e-05, "epoch": 0.7947598253275109, "percentage": 26.49, "elapsed_time": "14:46:09", "remaining_time": "1 day, 16:58:50"} +{"current_steps": 1825, "total_steps": 6870, "loss": 0.3359, "lr": 4.5942863861657394e-05, "epoch": 0.7969432314410481, "percentage": 26.56, "elapsed_time": "14:48:35", "remaining_time": "1 day, 16:56:25"} +{"current_steps": 1830, "total_steps": 6870, "loss": 0.3422, "lr": 4.590811149385348e-05, "epoch": 0.7991266375545851, "percentage": 26.64, "elapsed_time": "14:51:01", "remaining_time": "1 day, 16:53:59"} +{"current_steps": 1835, "total_steps": 6870, "loss": 0.3247, "lr": 4.5873224181348736e-05, "epoch": 0.8013100436681223, "percentage": 26.71, "elapsed_time": "14:53:27", "remaining_time": "1 day, 16:51:33"} +{"current_steps": 1840, "total_steps": 6870, "loss": 0.3304, "lr": 4.5838202149312145e-05, "epoch": 0.8034934497816594, "percentage": 26.78, "elapsed_time": "14:55:54", "remaining_time": "1 day, 16:49:07"} +{"current_steps": 1845, "total_steps": 6870, "loss": 0.3303, "lr": 4.580304562378218e-05, "epoch": 0.8056768558951966, "percentage": 26.86, "elapsed_time": "14:58:20", "remaining_time": "1 day, 16:46:41"} +{"current_steps": 1850, "total_steps": 6870, "loss": 0.3234, "lr": 4.576775483166536e-05, "epoch": 0.8078602620087336, "percentage": 26.93, "elapsed_time": "15:00:46", "remaining_time": "1 day, 16:44:15"} +{"current_steps": 1855, "total_steps": 6870, "loss": 0.3319, "lr": 4.5732330000734775e-05, "epoch": 0.8100436681222707, "percentage": 27.0, "elapsed_time": "15:03:12", "remaining_time": "1 day, 16:41:48"} +{"current_steps": 1860, "total_steps": 6870, "loss": 0.3278, "lr": 4.569677135962864e-05, "epoch": 0.8122270742358079, "percentage": 27.07, "elapsed_time": "15:05:38", "remaining_time": "1 day, 16:39:22"} +{"current_steps": 1865, "total_steps": 6870, "loss": 0.314, "lr": 4.5661079137848817e-05, "epoch": 0.8144104803493449, "percentage": 27.15, "elapsed_time": "15:08:04", "remaining_time": "1 day, 16:36:56"} +{"current_steps": 1870, "total_steps": 6870, "loss": 0.319, "lr": 4.562525356575928e-05, "epoch": 0.8165938864628821, "percentage": 27.22, "elapsed_time": "15:10:30", "remaining_time": "1 day, 16:34:30"} +{"current_steps": 1875, "total_steps": 6870, "loss": 0.3307, "lr": 4.558929487458472e-05, "epoch": 0.8187772925764192, "percentage": 27.29, "elapsed_time": "15:12:56", "remaining_time": "1 day, 16:32:04"} +{"current_steps": 1880, "total_steps": 6870, "loss": 0.3363, "lr": 4.5553203296408966e-05, "epoch": 0.8209606986899564, "percentage": 27.37, "elapsed_time": "15:15:22", "remaining_time": "1 day, 16:29:38"} +{"current_steps": 1885, "total_steps": 6870, "loss": 0.3228, "lr": 4.551697906417353e-05, "epoch": 0.8231441048034934, "percentage": 27.44, "elapsed_time": "15:17:48", "remaining_time": "1 day, 16:27:12"} +{"current_steps": 1890, "total_steps": 6870, "loss": 0.3371, "lr": 4.5480622411676105e-05, "epoch": 0.8253275109170306, "percentage": 27.51, "elapsed_time": "15:20:14", "remaining_time": "1 day, 16:24:46"} +{"current_steps": 1895, "total_steps": 6870, "loss": 0.3394, "lr": 4.5444133573569056e-05, "epoch": 0.8275109170305677, "percentage": 27.58, "elapsed_time": "15:22:40", "remaining_time": "1 day, 16:22:20"} +{"current_steps": 1900, "total_steps": 6870, "loss": 0.3413, "lr": 4.540751278535788e-05, "epoch": 0.8296943231441049, "percentage": 27.66, "elapsed_time": "15:25:06", "remaining_time": "1 day, 16:19:53"} +{"current_steps": 1905, "total_steps": 6870, "loss": 0.3237, "lr": 4.5370760283399714e-05, "epoch": 0.8318777292576419, "percentage": 27.73, "elapsed_time": "15:27:32", "remaining_time": "1 day, 16:17:27"} +{"current_steps": 1910, "total_steps": 6870, "loss": 0.3258, "lr": 4.533387630490179e-05, "epoch": 0.834061135371179, "percentage": 27.8, "elapsed_time": "15:29:58", "remaining_time": "1 day, 16:15:01"} +{"current_steps": 1915, "total_steps": 6870, "loss": 0.3365, "lr": 4.529686108791993e-05, "epoch": 0.8362445414847162, "percentage": 27.87, "elapsed_time": "15:32:24", "remaining_time": "1 day, 16:12:34"} +{"current_steps": 1920, "total_steps": 6870, "loss": 0.3261, "lr": 4.525971487135697e-05, "epoch": 0.8384279475982532, "percentage": 27.95, "elapsed_time": "15:34:50", "remaining_time": "1 day, 16:10:08"} +{"current_steps": 1925, "total_steps": 6870, "loss": 0.3314, "lr": 4.522243789496125e-05, "epoch": 0.8406113537117904, "percentage": 28.02, "elapsed_time": "15:37:16", "remaining_time": "1 day, 16:07:43"} +{"current_steps": 1930, "total_steps": 6870, "loss": 0.3294, "lr": 4.518503039932508e-05, "epoch": 0.8427947598253275, "percentage": 28.09, "elapsed_time": "15:39:43", "remaining_time": "1 day, 16:05:17"} +{"current_steps": 1935, "total_steps": 6870, "loss": 0.3346, "lr": 4.5147492625883133e-05, "epoch": 0.8449781659388647, "percentage": 28.17, "elapsed_time": "15:42:09", "remaining_time": "1 day, 16:02:51"} +{"current_steps": 1940, "total_steps": 6870, "loss": 0.3356, "lr": 4.510982481691093e-05, "epoch": 0.8471615720524017, "percentage": 28.24, "elapsed_time": "15:44:35", "remaining_time": "1 day, 16:00:26"} +{"current_steps": 1945, "total_steps": 6870, "loss": 0.328, "lr": 4.5072027215523256e-05, "epoch": 0.8493449781659389, "percentage": 28.31, "elapsed_time": "15:47:01", "remaining_time": "1 day, 15:58:00"} +{"current_steps": 1950, "total_steps": 6870, "loss": 0.3137, "lr": 4.503410006567262e-05, "epoch": 0.851528384279476, "percentage": 28.38, "elapsed_time": "15:49:28", "remaining_time": "1 day, 15:55:34"} +{"current_steps": 1955, "total_steps": 6870, "loss": 0.3367, "lr": 4.499604361214764e-05, "epoch": 0.8537117903930131, "percentage": 28.46, "elapsed_time": "15:51:54", "remaining_time": "1 day, 15:53:08"} +{"current_steps": 1960, "total_steps": 6870, "loss": 0.3305, "lr": 4.495785810057151e-05, "epoch": 0.8558951965065502, "percentage": 28.53, "elapsed_time": "15:54:20", "remaining_time": "1 day, 15:50:43"} +{"current_steps": 1965, "total_steps": 6870, "loss": 0.3148, "lr": 4.4919543777400354e-05, "epoch": 0.8580786026200873, "percentage": 28.6, "elapsed_time": "15:56:46", "remaining_time": "1 day, 15:48:17"} +{"current_steps": 1970, "total_steps": 6870, "loss": 0.3365, "lr": 4.488110088992168e-05, "epoch": 0.8602620087336245, "percentage": 28.68, "elapsed_time": "15:59:12", "remaining_time": "1 day, 15:45:51"} +{"current_steps": 1975, "total_steps": 6870, "loss": 0.3412, "lr": 4.4842529686252774e-05, "epoch": 0.8624454148471615, "percentage": 28.75, "elapsed_time": "16:01:38", "remaining_time": "1 day, 15:43:25"} +{"current_steps": 1980, "total_steps": 6870, "loss": 0.3319, "lr": 4.4803830415339124e-05, "epoch": 0.8646288209606987, "percentage": 28.82, "elapsed_time": "16:04:04", "remaining_time": "1 day, 15:40:59"} +{"current_steps": 1985, "total_steps": 6870, "loss": 0.3123, "lr": 4.476500332695275e-05, "epoch": 0.8668122270742358, "percentage": 28.89, "elapsed_time": "16:06:31", "remaining_time": "1 day, 15:38:33"} +{"current_steps": 1990, "total_steps": 6870, "loss": 0.3258, "lr": 4.472604867169063e-05, "epoch": 0.868995633187773, "percentage": 28.97, "elapsed_time": "16:08:57", "remaining_time": "1 day, 15:36:07"} +{"current_steps": 1995, "total_steps": 6870, "loss": 0.3321, "lr": 4.468696670097312e-05, "epoch": 0.87117903930131, "percentage": 29.04, "elapsed_time": "16:11:23", "remaining_time": "1 day, 15:33:41"} +{"current_steps": 2000, "total_steps": 6870, "loss": 0.339, "lr": 4.464775766704225e-05, "epoch": 0.8733624454148472, "percentage": 29.11, "elapsed_time": "16:13:49", "remaining_time": "1 day, 15:31:16"} +{"current_steps": 2005, "total_steps": 6870, "loss": 0.3232, "lr": 4.4608421822960165e-05, "epoch": 0.8755458515283843, "percentage": 29.18, "elapsed_time": "16:16:15", "remaining_time": "1 day, 15:28:50"} +{"current_steps": 2010, "total_steps": 6870, "loss": 0.3115, "lr": 4.456895942260745e-05, "epoch": 0.8777292576419214, "percentage": 29.26, "elapsed_time": "16:18:41", "remaining_time": "1 day, 15:26:24"} +{"current_steps": 2015, "total_steps": 6870, "loss": 0.3327, "lr": 4.4529370720681516e-05, "epoch": 0.8799126637554585, "percentage": 29.33, "elapsed_time": "16:21:07", "remaining_time": "1 day, 15:23:57"} +{"current_steps": 2020, "total_steps": 6870, "loss": 0.3234, "lr": 4.4489655972694946e-05, "epoch": 0.8820960698689956, "percentage": 29.4, "elapsed_time": "16:23:33", "remaining_time": "1 day, 15:21:31"} +{"current_steps": 2025, "total_steps": 6870, "loss": 0.3268, "lr": 4.444981543497384e-05, "epoch": 0.8842794759825328, "percentage": 29.48, "elapsed_time": "16:25:59", "remaining_time": "1 day, 15:19:05"} +{"current_steps": 2030, "total_steps": 6870, "loss": 0.3283, "lr": 4.440984936465618e-05, "epoch": 0.8864628820960698, "percentage": 29.55, "elapsed_time": "16:28:25", "remaining_time": "1 day, 15:16:39"} +{"current_steps": 2035, "total_steps": 6870, "loss": 0.3166, "lr": 4.436975801969013e-05, "epoch": 0.888646288209607, "percentage": 29.62, "elapsed_time": "16:30:51", "remaining_time": "1 day, 15:14:12"} +{"current_steps": 2040, "total_steps": 6870, "loss": 0.3275, "lr": 4.4329541658832454e-05, "epoch": 0.8908296943231441, "percentage": 29.69, "elapsed_time": "16:33:17", "remaining_time": "1 day, 15:11:46"} +{"current_steps": 2045, "total_steps": 6870, "loss": 0.3367, "lr": 4.428920054164673e-05, "epoch": 0.8930131004366813, "percentage": 29.77, "elapsed_time": "16:35:43", "remaining_time": "1 day, 15:09:20"} +{"current_steps": 2050, "total_steps": 6870, "loss": 0.3332, "lr": 4.424873492850177e-05, "epoch": 0.8951965065502183, "percentage": 29.84, "elapsed_time": "16:38:09", "remaining_time": "1 day, 15:06:54"} +{"current_steps": 2055, "total_steps": 6870, "loss": 0.3416, "lr": 4.42081450805699e-05, "epoch": 0.8973799126637555, "percentage": 29.91, "elapsed_time": "16:40:36", "remaining_time": "1 day, 15:04:28"} +{"current_steps": 2060, "total_steps": 6870, "loss": 0.3214, "lr": 4.4167431259825274e-05, "epoch": 0.8995633187772926, "percentage": 29.99, "elapsed_time": "16:43:02", "remaining_time": "1 day, 15:02:02"} +{"current_steps": 2065, "total_steps": 6870, "loss": 0.3403, "lr": 4.41265937290422e-05, "epoch": 0.9017467248908297, "percentage": 30.06, "elapsed_time": "16:45:28", "remaining_time": "1 day, 14:59:36"} +{"current_steps": 2070, "total_steps": 6870, "loss": 0.3161, "lr": 4.408563275179341e-05, "epoch": 0.9039301310043668, "percentage": 30.13, "elapsed_time": "16:47:54", "remaining_time": "1 day, 14:57:10"} +{"current_steps": 2075, "total_steps": 6870, "loss": 0.3513, "lr": 4.404454859244841e-05, "epoch": 0.9061135371179039, "percentage": 30.2, "elapsed_time": "16:50:20", "remaining_time": "1 day, 14:54:44"} +{"current_steps": 2080, "total_steps": 6870, "loss": 0.3551, "lr": 4.400334151617171e-05, "epoch": 0.9082969432314411, "percentage": 30.28, "elapsed_time": "16:52:46", "remaining_time": "1 day, 14:52:18"} +{"current_steps": 2085, "total_steps": 6870, "loss": 0.3202, "lr": 4.39620117889212e-05, "epoch": 0.9104803493449781, "percentage": 30.35, "elapsed_time": "16:55:12", "remaining_time": "1 day, 14:49:52"} +{"current_steps": 2090, "total_steps": 6870, "loss": 0.3264, "lr": 4.3920559677446335e-05, "epoch": 0.9126637554585153, "percentage": 30.42, "elapsed_time": "16:57:38", "remaining_time": "1 day, 14:47:26"} +{"current_steps": 2095, "total_steps": 6870, "loss": 0.3382, "lr": 4.387898544928648e-05, "epoch": 0.9148471615720524, "percentage": 30.49, "elapsed_time": "17:00:05", "remaining_time": "1 day, 14:45:00"} +{"current_steps": 2100, "total_steps": 6870, "loss": 0.3408, "lr": 4.383728937276915e-05, "epoch": 0.9170305676855895, "percentage": 30.57, "elapsed_time": "17:02:31", "remaining_time": "1 day, 14:42:34"} +{"current_steps": 2105, "total_steps": 6870, "loss": 0.3336, "lr": 4.37954717170083e-05, "epoch": 0.9192139737991266, "percentage": 30.64, "elapsed_time": "17:04:57", "remaining_time": "1 day, 14:40:09"} +{"current_steps": 2110, "total_steps": 6870, "loss": 0.3309, "lr": 4.375353275190259e-05, "epoch": 0.9213973799126638, "percentage": 30.71, "elapsed_time": "17:07:23", "remaining_time": "1 day, 14:37:43"} +{"current_steps": 2115, "total_steps": 6870, "loss": 0.3414, "lr": 4.371147274813362e-05, "epoch": 0.9235807860262009, "percentage": 30.79, "elapsed_time": "17:09:49", "remaining_time": "1 day, 14:35:17"} +{"current_steps": 2120, "total_steps": 6870, "loss": 0.3308, "lr": 4.3669291977164185e-05, "epoch": 0.925764192139738, "percentage": 30.86, "elapsed_time": "17:12:16", "remaining_time": "1 day, 14:32:52"} +{"current_steps": 2125, "total_steps": 6870, "loss": 0.3383, "lr": 4.362699071123655e-05, "epoch": 0.9279475982532751, "percentage": 30.93, "elapsed_time": "17:14:42", "remaining_time": "1 day, 14:30:26"} +{"current_steps": 2130, "total_steps": 6870, "loss": 0.3247, "lr": 4.358456922337068e-05, "epoch": 0.9301310043668122, "percentage": 31.0, "elapsed_time": "17:17:08", "remaining_time": "1 day, 14:28:00"} +{"current_steps": 2135, "total_steps": 6870, "loss": 0.3325, "lr": 4.354202778736245e-05, "epoch": 0.9323144104803494, "percentage": 31.08, "elapsed_time": "17:19:34", "remaining_time": "1 day, 14:25:34"} +{"current_steps": 2140, "total_steps": 6870, "loss": 0.3384, "lr": 4.3499366677781927e-05, "epoch": 0.9344978165938864, "percentage": 31.15, "elapsed_time": "17:22:01", "remaining_time": "1 day, 14:23:09"} +{"current_steps": 2145, "total_steps": 6870, "loss": 0.3155, "lr": 4.3456586169971555e-05, "epoch": 0.9366812227074236, "percentage": 31.22, "elapsed_time": "17:24:27", "remaining_time": "1 day, 14:20:43"} +{"current_steps": 2150, "total_steps": 6870, "loss": 0.3407, "lr": 4.341368654004442e-05, "epoch": 0.9388646288209607, "percentage": 31.3, "elapsed_time": "17:26:53", "remaining_time": "1 day, 14:18:16"} +{"current_steps": 2155, "total_steps": 6870, "loss": 0.3227, "lr": 4.3370668064882406e-05, "epoch": 0.9410480349344978, "percentage": 31.37, "elapsed_time": "17:29:19", "remaining_time": "1 day, 14:15:50"} +{"current_steps": 2160, "total_steps": 6870, "loss": 0.3488, "lr": 4.332753102213447e-05, "epoch": 0.9432314410480349, "percentage": 31.44, "elapsed_time": "17:31:45", "remaining_time": "1 day, 14:13:23"} +{"current_steps": 2165, "total_steps": 6870, "loss": 0.3237, "lr": 4.328427569021481e-05, "epoch": 0.9454148471615721, "percentage": 31.51, "elapsed_time": "17:34:11", "remaining_time": "1 day, 14:10:57"} +{"current_steps": 2170, "total_steps": 6870, "loss": 0.3137, "lr": 4.3240902348301134e-05, "epoch": 0.9475982532751092, "percentage": 31.59, "elapsed_time": "17:36:37", "remaining_time": "1 day, 14:08:31"} +{"current_steps": 2175, "total_steps": 6870, "loss": 0.3263, "lr": 4.319741127633273e-05, "epoch": 0.9497816593886463, "percentage": 31.66, "elapsed_time": "17:39:03", "remaining_time": "1 day, 14:06:05"} +{"current_steps": 2180, "total_steps": 6870, "loss": 0.3342, "lr": 4.315380275500879e-05, "epoch": 0.9519650655021834, "percentage": 31.73, "elapsed_time": "17:41:29", "remaining_time": "1 day, 14:03:40"} +{"current_steps": 2185, "total_steps": 6870, "loss": 0.3202, "lr": 4.311007706578654e-05, "epoch": 0.9541484716157205, "percentage": 31.8, "elapsed_time": "17:43:55", "remaining_time": "1 day, 14:01:14"} +{"current_steps": 2190, "total_steps": 6870, "loss": 0.3182, "lr": 4.306623449087939e-05, "epoch": 0.9563318777292577, "percentage": 31.88, "elapsed_time": "17:46:21", "remaining_time": "1 day, 13:58:48"} +{"current_steps": 2195, "total_steps": 6870, "loss": 0.3202, "lr": 4.3022275313255215e-05, "epoch": 0.9585152838427947, "percentage": 31.95, "elapsed_time": "17:48:47", "remaining_time": "1 day, 13:56:21"} +{"current_steps": 2200, "total_steps": 6870, "loss": 0.3134, "lr": 4.2978199816634404e-05, "epoch": 0.9606986899563319, "percentage": 32.02, "elapsed_time": "17:51:13", "remaining_time": "1 day, 13:53:55"} +{"current_steps": 2205, "total_steps": 6870, "loss": 0.3281, "lr": 4.293400828548812e-05, "epoch": 0.962882096069869, "percentage": 32.1, "elapsed_time": "17:53:39", "remaining_time": "1 day, 13:51:29"} +{"current_steps": 2210, "total_steps": 6870, "loss": 0.3257, "lr": 4.2889701005036444e-05, "epoch": 0.9650655021834061, "percentage": 32.17, "elapsed_time": "17:56:05", "remaining_time": "1 day, 13:49:03"} +{"current_steps": 2215, "total_steps": 6870, "loss": 0.3395, "lr": 4.28452782612465e-05, "epoch": 0.9672489082969432, "percentage": 32.24, "elapsed_time": "17:58:31", "remaining_time": "1 day, 13:46:37"} +{"current_steps": 2220, "total_steps": 6870, "loss": 0.3127, "lr": 4.2800740340830636e-05, "epoch": 0.9694323144104804, "percentage": 32.31, "elapsed_time": "18:00:58", "remaining_time": "1 day, 13:44:11"} +{"current_steps": 2225, "total_steps": 6870, "loss": 0.3139, "lr": 4.275608753124459e-05, "epoch": 0.9716157205240175, "percentage": 32.39, "elapsed_time": "18:03:24", "remaining_time": "1 day, 13:41:45"} +{"current_steps": 2230, "total_steps": 6870, "loss": 0.3144, "lr": 4.271132012068559e-05, "epoch": 0.9737991266375546, "percentage": 32.46, "elapsed_time": "18:05:50", "remaining_time": "1 day, 13:39:19"} +{"current_steps": 2235, "total_steps": 6870, "loss": 0.3217, "lr": 4.266643839809054e-05, "epoch": 0.9759825327510917, "percentage": 32.53, "elapsed_time": "18:08:16", "remaining_time": "1 day, 13:36:53"} +{"current_steps": 2240, "total_steps": 6870, "loss": 0.3263, "lr": 4.262144265313412e-05, "epoch": 0.9781659388646288, "percentage": 32.61, "elapsed_time": "18:10:42", "remaining_time": "1 day, 13:34:28"} +{"current_steps": 2245, "total_steps": 6870, "loss": 0.3174, "lr": 4.2576333176226944e-05, "epoch": 0.980349344978166, "percentage": 32.68, "elapsed_time": "18:13:08", "remaining_time": "1 day, 13:32:01"} +{"current_steps": 2250, "total_steps": 6870, "loss": 0.3324, "lr": 4.253111025851366e-05, "epoch": 0.982532751091703, "percentage": 32.75, "elapsed_time": "18:15:34", "remaining_time": "1 day, 13:29:34"} +{"current_steps": 2255, "total_steps": 6870, "loss": 0.3148, "lr": 4.2485774191871084e-05, "epoch": 0.9847161572052402, "percentage": 32.82, "elapsed_time": "18:18:00", "remaining_time": "1 day, 13:27:08"} +{"current_steps": 2260, "total_steps": 6870, "loss": 0.32, "lr": 4.244032526890633e-05, "epoch": 0.9868995633187773, "percentage": 32.9, "elapsed_time": "18:20:26", "remaining_time": "1 day, 13:24:42"} +{"current_steps": 2265, "total_steps": 6870, "loss": 0.3208, "lr": 4.239476378295489e-05, "epoch": 0.9890829694323144, "percentage": 32.97, "elapsed_time": "18:22:52", "remaining_time": "1 day, 13:22:16"} +{"current_steps": 2270, "total_steps": 6870, "loss": 0.3108, "lr": 4.2349090028078766e-05, "epoch": 0.9912663755458515, "percentage": 33.04, "elapsed_time": "18:25:18", "remaining_time": "1 day, 13:19:50"} +{"current_steps": 2275, "total_steps": 6870, "loss": 0.3114, "lr": 4.230330429906456e-05, "epoch": 0.9934497816593887, "percentage": 33.11, "elapsed_time": "18:27:45", "remaining_time": "1 day, 13:17:25"} +{"current_steps": 2280, "total_steps": 6870, "loss": 0.3168, "lr": 4.225740689142158e-05, "epoch": 0.9956331877729258, "percentage": 33.19, "elapsed_time": "18:30:11", "remaining_time": "1 day, 13:14:59"} +{"current_steps": 2285, "total_steps": 6870, "loss": 0.3174, "lr": 4.221139810137992e-05, "epoch": 0.9978165938864629, "percentage": 33.26, "elapsed_time": "18:32:37", "remaining_time": "1 day, 13:12:33"} +{"current_steps": 2290, "total_steps": 6870, "loss": 0.3168, "lr": 4.216527822588857e-05, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "18:35:03", "remaining_time": "1 day, 13:10:07"} +{"current_steps": 2295, "total_steps": 6870, "loss": 0.2941, "lr": 4.2119047562613476e-05, "epoch": 1.002183406113537, "percentage": 33.41, "elapsed_time": "18:37:29", "remaining_time": "1 day, 13:07:41"} +{"current_steps": 2300, "total_steps": 6870, "loss": 0.2787, "lr": 4.207270640993561e-05, "epoch": 1.0043668122270741, "percentage": 33.48, "elapsed_time": "18:39:56", "remaining_time": "1 day, 13:05:15"} +{"current_steps": 2305, "total_steps": 6870, "loss": 0.2686, "lr": 4.202625506694911e-05, "epoch": 1.0065502183406114, "percentage": 33.55, "elapsed_time": "18:42:22", "remaining_time": "1 day, 13:02:49"} +{"current_steps": 2310, "total_steps": 6870, "loss": 0.2809, "lr": 4.197969383345925e-05, "epoch": 1.0087336244541485, "percentage": 33.62, "elapsed_time": "18:44:48", "remaining_time": "1 day, 13:00:23"} +{"current_steps": 2315, "total_steps": 6870, "loss": 0.2979, "lr": 4.19330230099806e-05, "epoch": 1.0109170305676856, "percentage": 33.7, "elapsed_time": "18:47:14", "remaining_time": "1 day, 12:57:57"} +{"current_steps": 2320, "total_steps": 6870, "loss": 0.2967, "lr": 4.188624289773501e-05, "epoch": 1.0131004366812226, "percentage": 33.77, "elapsed_time": "18:49:40", "remaining_time": "1 day, 12:55:32"} +{"current_steps": 2325, "total_steps": 6870, "loss": 0.2758, "lr": 4.1839353798649717e-05, "epoch": 1.01528384279476, "percentage": 33.84, "elapsed_time": "18:52:07", "remaining_time": "1 day, 12:53:06"} +{"current_steps": 2330, "total_steps": 6870, "loss": 0.2766, "lr": 4.1792356015355375e-05, "epoch": 1.017467248908297, "percentage": 33.92, "elapsed_time": "18:54:33", "remaining_time": "1 day, 12:50:40"} +{"current_steps": 2335, "total_steps": 6870, "loss": 0.286, "lr": 4.17452498511841e-05, "epoch": 1.019650655021834, "percentage": 33.99, "elapsed_time": "18:56:59", "remaining_time": "1 day, 12:48:14"} +{"current_steps": 2340, "total_steps": 6870, "loss": 0.2856, "lr": 4.169803561016754e-05, "epoch": 1.0218340611353711, "percentage": 34.06, "elapsed_time": "18:59:25", "remaining_time": "1 day, 12:45:48"} +{"current_steps": 2345, "total_steps": 6870, "loss": 0.287, "lr": 4.1650713597034845e-05, "epoch": 1.0240174672489082, "percentage": 34.13, "elapsed_time": "19:01:51", "remaining_time": "1 day, 12:43:22"} +{"current_steps": 2350, "total_steps": 6870, "loss": 0.2927, "lr": 4.16032841172108e-05, "epoch": 1.0262008733624455, "percentage": 34.21, "elapsed_time": "19:04:17", "remaining_time": "1 day, 12:40:56"} +{"current_steps": 2355, "total_steps": 6870, "loss": 0.2858, "lr": 4.155574747681375e-05, "epoch": 1.0283842794759825, "percentage": 34.28, "elapsed_time": "19:06:43", "remaining_time": "1 day, 12:38:30"} +{"current_steps": 2360, "total_steps": 6870, "loss": 0.2991, "lr": 4.1508103982653715e-05, "epoch": 1.0305676855895196, "percentage": 34.35, "elapsed_time": "19:09:10", "remaining_time": "1 day, 12:36:04"} +{"current_steps": 2365, "total_steps": 6870, "loss": 0.2776, "lr": 4.146035394223034e-05, "epoch": 1.0327510917030567, "percentage": 34.43, "elapsed_time": "19:11:36", "remaining_time": "1 day, 12:33:38"} +{"current_steps": 2370, "total_steps": 6870, "loss": 0.2748, "lr": 4.1412497663730954e-05, "epoch": 1.034934497816594, "percentage": 34.5, "elapsed_time": "19:14:02", "remaining_time": "1 day, 12:31:12"} +{"current_steps": 2375, "total_steps": 6870, "loss": 0.2841, "lr": 4.1364535456028555e-05, "epoch": 1.037117903930131, "percentage": 34.57, "elapsed_time": "19:16:28", "remaining_time": "1 day, 12:28:46"} +{"current_steps": 2380, "total_steps": 6870, "loss": 0.2619, "lr": 4.1316467628679834e-05, "epoch": 1.039301310043668, "percentage": 34.64, "elapsed_time": "19:18:54", "remaining_time": "1 day, 12:26:19"} +{"current_steps": 2385, "total_steps": 6870, "loss": 0.2857, "lr": 4.126829449192316e-05, "epoch": 1.0414847161572052, "percentage": 34.72, "elapsed_time": "19:21:20", "remaining_time": "1 day, 12:23:53"} +{"current_steps": 2390, "total_steps": 6870, "loss": 0.2787, "lr": 4.1220016356676615e-05, "epoch": 1.0436681222707425, "percentage": 34.79, "elapsed_time": "19:23:46", "remaining_time": "1 day, 12:21:27"} +{"current_steps": 2395, "total_steps": 6870, "loss": 0.2824, "lr": 4.117163353453593e-05, "epoch": 1.0458515283842795, "percentage": 34.86, "elapsed_time": "19:26:12", "remaining_time": "1 day, 12:19:01"} +{"current_steps": 2400, "total_steps": 6870, "loss": 0.2831, "lr": 4.112314633777251e-05, "epoch": 1.0480349344978166, "percentage": 34.93, "elapsed_time": "19:28:38", "remaining_time": "1 day, 12:16:35"} +{"current_steps": 2405, "total_steps": 6870, "loss": 0.2741, "lr": 4.107455507933144e-05, "epoch": 1.0502183406113537, "percentage": 35.01, "elapsed_time": "19:31:04", "remaining_time": "1 day, 12:14:09"} +{"current_steps": 2410, "total_steps": 6870, "loss": 0.2784, "lr": 4.102586007282938e-05, "epoch": 1.0524017467248907, "percentage": 35.08, "elapsed_time": "19:33:30", "remaining_time": "1 day, 12:11:44"} +{"current_steps": 2415, "total_steps": 6870, "loss": 0.2822, "lr": 4.0977061632552656e-05, "epoch": 1.054585152838428, "percentage": 35.15, "elapsed_time": "19:35:57", "remaining_time": "1 day, 12:09:18"} +{"current_steps": 2420, "total_steps": 6870, "loss": 0.2918, "lr": 4.092816007345515e-05, "epoch": 1.056768558951965, "percentage": 35.23, "elapsed_time": "19:38:23", "remaining_time": "1 day, 12:06:52"} +{"current_steps": 2425, "total_steps": 6870, "loss": 0.2657, "lr": 4.087915571115629e-05, "epoch": 1.0589519650655022, "percentage": 35.3, "elapsed_time": "19:40:49", "remaining_time": "1 day, 12:04:26"} +{"current_steps": 2430, "total_steps": 6870, "loss": 0.2821, "lr": 4.083004886193902e-05, "epoch": 1.0611353711790392, "percentage": 35.37, "elapsed_time": "19:43:15", "remaining_time": "1 day, 12:02:00"} +{"current_steps": 2435, "total_steps": 6870, "loss": 0.2876, "lr": 4.0780839842747754e-05, "epoch": 1.0633187772925765, "percentage": 35.44, "elapsed_time": "19:45:41", "remaining_time": "1 day, 11:59:34"} +{"current_steps": 2440, "total_steps": 6870, "loss": 0.2902, "lr": 4.073152897118632e-05, "epoch": 1.0655021834061136, "percentage": 35.52, "elapsed_time": "19:48:08", "remaining_time": "1 day, 11:57:08"} +{"current_steps": 2445, "total_steps": 6870, "loss": 0.294, "lr": 4.0682116565515935e-05, "epoch": 1.0676855895196506, "percentage": 35.59, "elapsed_time": "19:50:34", "remaining_time": "1 day, 11:54:42"} +{"current_steps": 2450, "total_steps": 6870, "loss": 0.2754, "lr": 4.063260294465311e-05, "epoch": 1.0698689956331877, "percentage": 35.66, "elapsed_time": "19:53:00", "remaining_time": "1 day, 11:52:16"} +{"current_steps": 2455, "total_steps": 6870, "loss": 0.2756, "lr": 4.058298842816765e-05, "epoch": 1.072052401746725, "percentage": 35.74, "elapsed_time": "19:55:26", "remaining_time": "1 day, 11:49:50"} +{"current_steps": 2460, "total_steps": 6870, "loss": 0.2749, "lr": 4.0533273336280516e-05, "epoch": 1.074235807860262, "percentage": 35.81, "elapsed_time": "19:57:52", "remaining_time": "1 day, 11:47:24"} +{"current_steps": 2465, "total_steps": 6870, "loss": 0.2937, "lr": 4.048345798986186e-05, "epoch": 1.0764192139737991, "percentage": 35.88, "elapsed_time": "20:00:18", "remaining_time": "1 day, 11:44:58"} +{"current_steps": 2470, "total_steps": 6870, "loss": 0.2772, "lr": 4.043354271042884e-05, "epoch": 1.0786026200873362, "percentage": 35.95, "elapsed_time": "20:02:44", "remaining_time": "1 day, 11:42:32"} +{"current_steps": 2475, "total_steps": 6870, "loss": 0.2783, "lr": 4.038352782014362e-05, "epoch": 1.0807860262008733, "percentage": 36.03, "elapsed_time": "20:05:10", "remaining_time": "1 day, 11:40:06"} +{"current_steps": 2480, "total_steps": 6870, "loss": 0.2845, "lr": 4.033341364181128e-05, "epoch": 1.0829694323144106, "percentage": 36.1, "elapsed_time": "20:07:37", "remaining_time": "1 day, 11:37:40"} +{"current_steps": 2485, "total_steps": 6870, "loss": 0.2815, "lr": 4.028320049887771e-05, "epoch": 1.0851528384279476, "percentage": 36.17, "elapsed_time": "20:10:03", "remaining_time": "1 day, 11:35:14"} +{"current_steps": 2490, "total_steps": 6870, "loss": 0.2767, "lr": 4.023288871542752e-05, "epoch": 1.0873362445414847, "percentage": 36.24, "elapsed_time": "20:12:29", "remaining_time": "1 day, 11:32:48"} +{"current_steps": 2495, "total_steps": 6870, "loss": 0.2802, "lr": 4.0182478616182015e-05, "epoch": 1.0895196506550218, "percentage": 36.32, "elapsed_time": "20:14:55", "remaining_time": "1 day, 11:30:22"} +{"current_steps": 2500, "total_steps": 6870, "loss": 0.2885, "lr": 4.013197052649699e-05, "epoch": 1.091703056768559, "percentage": 36.39, "elapsed_time": "20:17:21", "remaining_time": "1 day, 11:27:56"} +{"current_steps": 2505, "total_steps": 6870, "loss": 0.2849, "lr": 4.0081364772360725e-05, "epoch": 1.0938864628820961, "percentage": 36.46, "elapsed_time": "20:19:47", "remaining_time": "1 day, 11:25:30"} +{"current_steps": 2510, "total_steps": 6870, "loss": 0.2748, "lr": 4.003066168039183e-05, "epoch": 1.0960698689956332, "percentage": 36.54, "elapsed_time": "20:22:13", "remaining_time": "1 day, 11:23:04"} +{"current_steps": 2515, "total_steps": 6870, "loss": 0.2783, "lr": 3.9979861577837155e-05, "epoch": 1.0982532751091703, "percentage": 36.61, "elapsed_time": "20:24:40", "remaining_time": "1 day, 11:20:38"} +{"current_steps": 2520, "total_steps": 6870, "loss": 0.2811, "lr": 3.9928964792569655e-05, "epoch": 1.1004366812227073, "percentage": 36.68, "elapsed_time": "20:27:06", "remaining_time": "1 day, 11:18:12"} +{"current_steps": 2525, "total_steps": 6870, "loss": 0.2805, "lr": 3.987797165308633e-05, "epoch": 1.1026200873362446, "percentage": 36.75, "elapsed_time": "20:29:32", "remaining_time": "1 day, 11:15:46"} +{"current_steps": 2530, "total_steps": 6870, "loss": 0.2929, "lr": 3.982688248850604e-05, "epoch": 1.1048034934497817, "percentage": 36.83, "elapsed_time": "20:31:58", "remaining_time": "1 day, 11:13:20"} +{"current_steps": 2535, "total_steps": 6870, "loss": 0.2913, "lr": 3.977569762856741e-05, "epoch": 1.1069868995633187, "percentage": 36.9, "elapsed_time": "20:34:24", "remaining_time": "1 day, 11:10:54"} +{"current_steps": 2540, "total_steps": 6870, "loss": 0.2834, "lr": 3.9724417403626705e-05, "epoch": 1.1091703056768558, "percentage": 36.97, "elapsed_time": "20:36:50", "remaining_time": "1 day, 11:08:28"} +{"current_steps": 2545, "total_steps": 6870, "loss": 0.2927, "lr": 3.9673042144655697e-05, "epoch": 1.111353711790393, "percentage": 37.05, "elapsed_time": "20:39:16", "remaining_time": "1 day, 11:06:02"} +{"current_steps": 2550, "total_steps": 6870, "loss": 0.2778, "lr": 3.9621572183239515e-05, "epoch": 1.1135371179039302, "percentage": 37.12, "elapsed_time": "20:41:42", "remaining_time": "1 day, 11:03:36"} +{"current_steps": 2555, "total_steps": 6870, "loss": 0.3014, "lr": 3.957000785157452e-05, "epoch": 1.1157205240174672, "percentage": 37.19, "elapsed_time": "20:44:09", "remaining_time": "1 day, 11:01:10"} +{"current_steps": 2560, "total_steps": 6870, "loss": 0.268, "lr": 3.951834948246616e-05, "epoch": 1.1179039301310043, "percentage": 37.26, "elapsed_time": "20:46:35", "remaining_time": "1 day, 10:58:44"} +{"current_steps": 2565, "total_steps": 6870, "loss": 0.2999, "lr": 3.946659740932681e-05, "epoch": 1.1200873362445414, "percentage": 37.34, "elapsed_time": "20:49:01", "remaining_time": "1 day, 10:56:18"} +{"current_steps": 2570, "total_steps": 6870, "loss": 0.2788, "lr": 3.9414751966173634e-05, "epoch": 1.1222707423580787, "percentage": 37.41, "elapsed_time": "20:51:27", "remaining_time": "1 day, 10:53:52"} +{"current_steps": 2575, "total_steps": 6870, "loss": 0.2764, "lr": 3.93628134876264e-05, "epoch": 1.1244541484716157, "percentage": 37.48, "elapsed_time": "20:53:53", "remaining_time": "1 day, 10:51:26"} +{"current_steps": 2580, "total_steps": 6870, "loss": 0.2809, "lr": 3.9310782308905395e-05, "epoch": 1.1266375545851528, "percentage": 37.55, "elapsed_time": "20:56:19", "remaining_time": "1 day, 10:49:00"} +{"current_steps": 2585, "total_steps": 6870, "loss": 0.2668, "lr": 3.925865876582915e-05, "epoch": 1.1288209606986899, "percentage": 37.63, "elapsed_time": "20:58:46", "remaining_time": "1 day, 10:46:34"} +{"current_steps": 2590, "total_steps": 6870, "loss": 0.2762, "lr": 3.920644319481237e-05, "epoch": 1.1310043668122272, "percentage": 37.7, "elapsed_time": "21:01:12", "remaining_time": "1 day, 10:44:09"} +{"current_steps": 2595, "total_steps": 6870, "loss": 0.2828, "lr": 3.915413593286371e-05, "epoch": 1.1331877729257642, "percentage": 37.77, "elapsed_time": "21:03:38", "remaining_time": "1 day, 10:41:43"} +{"current_steps": 2600, "total_steps": 6870, "loss": 0.2752, "lr": 3.910173731758362e-05, "epoch": 1.1353711790393013, "percentage": 37.85, "elapsed_time": "21:06:04", "remaining_time": "1 day, 10:39:17"} +{"current_steps": 2605, "total_steps": 6870, "loss": 0.2764, "lr": 3.904924768716216e-05, "epoch": 1.1375545851528384, "percentage": 37.92, "elapsed_time": "21:08:30", "remaining_time": "1 day, 10:36:50"} +{"current_steps": 2610, "total_steps": 6870, "loss": 0.2746, "lr": 3.89966673803768e-05, "epoch": 1.1397379912663754, "percentage": 37.99, "elapsed_time": "21:10:56", "remaining_time": "1 day, 10:34:24"} +{"current_steps": 2615, "total_steps": 6870, "loss": 0.2911, "lr": 3.894399673659031e-05, "epoch": 1.1419213973799127, "percentage": 38.06, "elapsed_time": "21:13:22", "remaining_time": "1 day, 10:31:58"} +{"current_steps": 2620, "total_steps": 6870, "loss": 0.2687, "lr": 3.889123609574843e-05, "epoch": 1.1441048034934498, "percentage": 38.14, "elapsed_time": "21:15:48", "remaining_time": "1 day, 10:29:32"} +{"current_steps": 2625, "total_steps": 6870, "loss": 0.29, "lr": 3.883838579837781e-05, "epoch": 1.1462882096069869, "percentage": 38.21, "elapsed_time": "21:18:15", "remaining_time": "1 day, 10:27:06"} +{"current_steps": 2630, "total_steps": 6870, "loss": 0.2852, "lr": 3.8785446185583755e-05, "epoch": 1.1484716157205241, "percentage": 38.28, "elapsed_time": "21:20:41", "remaining_time": "1 day, 10:24:40"} +{"current_steps": 2635, "total_steps": 6870, "loss": 0.2844, "lr": 3.873241759904801e-05, "epoch": 1.1506550218340612, "percentage": 38.36, "elapsed_time": "21:23:07", "remaining_time": "1 day, 10:22:14"} +{"current_steps": 2640, "total_steps": 6870, "loss": 0.2926, "lr": 3.8679300381026585e-05, "epoch": 1.1528384279475983, "percentage": 38.43, "elapsed_time": "21:25:33", "remaining_time": "1 day, 10:19:48"} +{"current_steps": 2645, "total_steps": 6870, "loss": 0.2701, "lr": 3.862609487434752e-05, "epoch": 1.1550218340611353, "percentage": 38.5, "elapsed_time": "21:27:59", "remaining_time": "1 day, 10:17:22"} +{"current_steps": 2650, "total_steps": 6870, "loss": 0.2813, "lr": 3.857280142240871e-05, "epoch": 1.1572052401746724, "percentage": 38.57, "elapsed_time": "21:30:25", "remaining_time": "1 day, 10:14:56"} +{"current_steps": 2655, "total_steps": 6870, "loss": 0.2989, "lr": 3.851942036917565e-05, "epoch": 1.1593886462882097, "percentage": 38.65, "elapsed_time": "21:32:51", "remaining_time": "1 day, 10:12:30"} +{"current_steps": 2660, "total_steps": 6870, "loss": 0.2831, "lr": 3.846595205917922e-05, "epoch": 1.1615720524017468, "percentage": 38.72, "elapsed_time": "21:35:17", "remaining_time": "1 day, 10:10:04"} +{"current_steps": 2665, "total_steps": 6870, "loss": 0.2838, "lr": 3.8412396837513484e-05, "epoch": 1.1637554585152838, "percentage": 38.79, "elapsed_time": "21:37:44", "remaining_time": "1 day, 10:07:38"} +{"current_steps": 2670, "total_steps": 6870, "loss": 0.2896, "lr": 3.835875504983344e-05, "epoch": 1.165938864628821, "percentage": 38.86, "elapsed_time": "21:40:10", "remaining_time": "1 day, 10:05:12"} +{"current_steps": 2675, "total_steps": 6870, "loss": 0.281, "lr": 3.830502704235282e-05, "epoch": 1.1681222707423582, "percentage": 38.94, "elapsed_time": "21:42:36", "remaining_time": "1 day, 10:02:46"} +{"current_steps": 2680, "total_steps": 6870, "loss": 0.2883, "lr": 3.825121316184181e-05, "epoch": 1.1703056768558953, "percentage": 39.01, "elapsed_time": "21:45:01", "remaining_time": "1 day, 10:00:19"} +{"current_steps": 2685, "total_steps": 6870, "loss": 0.266, "lr": 3.8197313755624855e-05, "epoch": 1.1724890829694323, "percentage": 39.08, "elapsed_time": "21:47:27", "remaining_time": "1 day, 9:57:53"} +{"current_steps": 2690, "total_steps": 6870, "loss": 0.2792, "lr": 3.8143329171578366e-05, "epoch": 1.1746724890829694, "percentage": 39.16, "elapsed_time": "21:49:53", "remaining_time": "1 day, 9:55:27"} +{"current_steps": 2695, "total_steps": 6870, "loss": 0.2994, "lr": 3.808925975812855e-05, "epoch": 1.1768558951965065, "percentage": 39.23, "elapsed_time": "21:52:19", "remaining_time": "1 day, 9:53:01"} +{"current_steps": 2700, "total_steps": 6870, "loss": 0.2909, "lr": 3.803510586424908e-05, "epoch": 1.1790393013100438, "percentage": 39.3, "elapsed_time": "21:54:46", "remaining_time": "1 day, 9:50:35"} +{"current_steps": 2705, "total_steps": 6870, "loss": 0.2829, "lr": 3.798086783945892e-05, "epoch": 1.1812227074235808, "percentage": 39.37, "elapsed_time": "21:57:12", "remaining_time": "1 day, 9:48:08"} +{"current_steps": 2710, "total_steps": 6870, "loss": 0.2843, "lr": 3.792654603381998e-05, "epoch": 1.1834061135371179, "percentage": 39.45, "elapsed_time": "21:59:37", "remaining_time": "1 day, 9:45:42"} +{"current_steps": 2715, "total_steps": 6870, "loss": 0.2719, "lr": 3.7872140797934956e-05, "epoch": 1.185589519650655, "percentage": 39.52, "elapsed_time": "22:02:03", "remaining_time": "1 day, 9:43:16"} +{"current_steps": 2720, "total_steps": 6870, "loss": 0.2759, "lr": 3.781765248294499e-05, "epoch": 1.1877729257641922, "percentage": 39.59, "elapsed_time": "22:04:29", "remaining_time": "1 day, 9:40:50"} +{"current_steps": 2725, "total_steps": 6870, "loss": 0.2857, "lr": 3.7763081440527435e-05, "epoch": 1.1899563318777293, "percentage": 39.67, "elapsed_time": "22:06:56", "remaining_time": "1 day, 9:38:24"} +{"current_steps": 2730, "total_steps": 6870, "loss": 0.2863, "lr": 3.7708428022893596e-05, "epoch": 1.1921397379912664, "percentage": 39.74, "elapsed_time": "22:09:22", "remaining_time": "1 day, 9:35:57"} +{"current_steps": 2735, "total_steps": 6870, "loss": 0.272, "lr": 3.7653692582786416e-05, "epoch": 1.1943231441048034, "percentage": 39.81, "elapsed_time": "22:11:48", "remaining_time": "1 day, 9:33:31"} +{"current_steps": 2740, "total_steps": 6870, "loss": 0.2715, "lr": 3.7598875473478246e-05, "epoch": 1.1965065502183405, "percentage": 39.88, "elapsed_time": "22:14:14", "remaining_time": "1 day, 9:31:05"} +{"current_steps": 2745, "total_steps": 6870, "loss": 0.271, "lr": 3.7543977048768534e-05, "epoch": 1.1986899563318778, "percentage": 39.96, "elapsed_time": "22:16:40", "remaining_time": "1 day, 9:28:39"} +{"current_steps": 2750, "total_steps": 6870, "loss": 0.2845, "lr": 3.748899766298157e-05, "epoch": 1.2008733624454149, "percentage": 40.03, "elapsed_time": "22:19:06", "remaining_time": "1 day, 9:26:13"} +{"current_steps": 2755, "total_steps": 6870, "loss": 0.2755, "lr": 3.743393767096416e-05, "epoch": 1.203056768558952, "percentage": 40.1, "elapsed_time": "22:21:32", "remaining_time": "1 day, 9:23:47"} +{"current_steps": 2760, "total_steps": 6870, "loss": 0.2793, "lr": 3.737879742808337e-05, "epoch": 1.205240174672489, "percentage": 40.17, "elapsed_time": "22:23:58", "remaining_time": "1 day, 9:21:21"} +{"current_steps": 2765, "total_steps": 6870, "loss": 0.2877, "lr": 3.732357729022421e-05, "epoch": 1.2074235807860263, "percentage": 40.25, "elapsed_time": "22:26:24", "remaining_time": "1 day, 9:18:55"} +{"current_steps": 2770, "total_steps": 6870, "loss": 0.2869, "lr": 3.7268277613787356e-05, "epoch": 1.2096069868995634, "percentage": 40.32, "elapsed_time": "22:28:51", "remaining_time": "1 day, 9:16:29"} +{"current_steps": 2775, "total_steps": 6870, "loss": 0.2853, "lr": 3.721289875568683e-05, "epoch": 1.2117903930131004, "percentage": 40.39, "elapsed_time": "22:31:17", "remaining_time": "1 day, 9:14:04"} +{"current_steps": 2780, "total_steps": 6870, "loss": 0.2699, "lr": 3.715744107334773e-05, "epoch": 1.2139737991266375, "percentage": 40.47, "elapsed_time": "22:33:43", "remaining_time": "1 day, 9:11:38"} +{"current_steps": 2785, "total_steps": 6870, "loss": 0.2823, "lr": 3.7101904924703865e-05, "epoch": 1.2161572052401746, "percentage": 40.54, "elapsed_time": "22:36:09", "remaining_time": "1 day, 9:09:11"} +{"current_steps": 2790, "total_steps": 6870, "loss": 0.2771, "lr": 3.704629066819549e-05, "epoch": 1.2183406113537119, "percentage": 40.61, "elapsed_time": "22:38:35", "remaining_time": "1 day, 9:06:45"} +{"current_steps": 2795, "total_steps": 6870, "loss": 0.2699, "lr": 3.6990598662767014e-05, "epoch": 1.220524017467249, "percentage": 40.68, "elapsed_time": "22:41:01", "remaining_time": "1 day, 9:04:19"} +{"current_steps": 2800, "total_steps": 6870, "loss": 0.2726, "lr": 3.693482926786461e-05, "epoch": 1.222707423580786, "percentage": 40.76, "elapsed_time": "22:43:27", "remaining_time": "1 day, 9:01:53"} +{"current_steps": 2805, "total_steps": 6870, "loss": 0.2891, "lr": 3.687898284343396e-05, "epoch": 1.224890829694323, "percentage": 40.83, "elapsed_time": "22:45:53", "remaining_time": "1 day, 8:59:27"} +{"current_steps": 2810, "total_steps": 6870, "loss": 0.2805, "lr": 3.68230597499179e-05, "epoch": 1.2270742358078603, "percentage": 40.9, "elapsed_time": "22:48:19", "remaining_time": "1 day, 8:57:01"} +{"current_steps": 2815, "total_steps": 6870, "loss": 0.3019, "lr": 3.676706034825409e-05, "epoch": 1.2292576419213974, "percentage": 40.98, "elapsed_time": "22:50:46", "remaining_time": "1 day, 8:54:35"} +{"current_steps": 2820, "total_steps": 6870, "loss": 0.2748, "lr": 3.6710984999872735e-05, "epoch": 1.2314410480349345, "percentage": 41.05, "elapsed_time": "22:53:12", "remaining_time": "1 day, 8:52:09"} +{"current_steps": 2825, "total_steps": 6870, "loss": 0.2904, "lr": 3.665483406669418e-05, "epoch": 1.2336244541484715, "percentage": 41.12, "elapsed_time": "22:55:38", "remaining_time": "1 day, 8:49:43"} +{"current_steps": 2830, "total_steps": 6870, "loss": 0.2856, "lr": 3.6598607911126614e-05, "epoch": 1.2358078602620086, "percentage": 41.19, "elapsed_time": "22:58:04", "remaining_time": "1 day, 8:47:17"} +{"current_steps": 2835, "total_steps": 6870, "loss": 0.2659, "lr": 3.654230689606372e-05, "epoch": 1.237991266375546, "percentage": 41.27, "elapsed_time": "23:00:30", "remaining_time": "1 day, 8:44:51"} +{"current_steps": 2840, "total_steps": 6870, "loss": 0.278, "lr": 3.648593138488235e-05, "epoch": 1.240174672489083, "percentage": 41.34, "elapsed_time": "23:02:56", "remaining_time": "1 day, 8:42:25"} +{"current_steps": 2845, "total_steps": 6870, "loss": 0.2804, "lr": 3.6429481741440176e-05, "epoch": 1.24235807860262, "percentage": 41.41, "elapsed_time": "23:05:22", "remaining_time": "1 day, 8:39:59"} +{"current_steps": 2850, "total_steps": 6870, "loss": 0.278, "lr": 3.63729583300733e-05, "epoch": 1.244541484716157, "percentage": 41.48, "elapsed_time": "23:07:49", "remaining_time": "1 day, 8:37:33"} +{"current_steps": 2855, "total_steps": 6870, "loss": 0.2874, "lr": 3.6316361515593966e-05, "epoch": 1.2467248908296944, "percentage": 41.56, "elapsed_time": "23:10:15", "remaining_time": "1 day, 8:35:07"} +{"current_steps": 2860, "total_steps": 6870, "loss": 0.276, "lr": 3.625969166328815e-05, "epoch": 1.2489082969432315, "percentage": 41.63, "elapsed_time": "23:12:41", "remaining_time": "1 day, 8:32:41"} +{"current_steps": 2865, "total_steps": 6870, "loss": 0.2884, "lr": 3.620294913891326e-05, "epoch": 1.2510917030567685, "percentage": 41.7, "elapsed_time": "23:15:07", "remaining_time": "1 day, 8:30:15"} +{"current_steps": 2870, "total_steps": 6870, "loss": 0.2859, "lr": 3.614613430869571e-05, "epoch": 1.2532751091703056, "percentage": 41.78, "elapsed_time": "23:17:33", "remaining_time": "1 day, 8:27:49"} +{"current_steps": 2875, "total_steps": 6870, "loss": 0.2884, "lr": 3.608924753932862e-05, "epoch": 1.2554585152838427, "percentage": 41.85, "elapsed_time": "23:19:59", "remaining_time": "1 day, 8:25:23"} +{"current_steps": 2880, "total_steps": 6870, "loss": 0.2858, "lr": 3.6032289197969374e-05, "epoch": 1.25764192139738, "percentage": 41.92, "elapsed_time": "23:22:25", "remaining_time": "1 day, 8:22:56"} +{"current_steps": 2885, "total_steps": 6870, "loss": 0.287, "lr": 3.5975259652237333e-05, "epoch": 1.259825327510917, "percentage": 41.99, "elapsed_time": "23:24:51", "remaining_time": "1 day, 8:20:30"} +{"current_steps": 2890, "total_steps": 6870, "loss": 0.2734, "lr": 3.5918159270211424e-05, "epoch": 1.262008733624454, "percentage": 42.07, "elapsed_time": "23:27:17", "remaining_time": "1 day, 8:18:04"} +{"current_steps": 2895, "total_steps": 6870, "loss": 0.2774, "lr": 3.586098842042773e-05, "epoch": 1.2641921397379914, "percentage": 42.14, "elapsed_time": "23:29:44", "remaining_time": "1 day, 8:15:38"} +{"current_steps": 2900, "total_steps": 6870, "loss": 0.2775, "lr": 3.580374747187719e-05, "epoch": 1.2663755458515285, "percentage": 42.21, "elapsed_time": "23:32:10", "remaining_time": "1 day, 8:13:12"} +{"current_steps": 2905, "total_steps": 6870, "loss": 0.2747, "lr": 3.574643679400313e-05, "epoch": 1.2685589519650655, "percentage": 42.29, "elapsed_time": "23:34:36", "remaining_time": "1 day, 8:10:46"} +{"current_steps": 2910, "total_steps": 6870, "loss": 0.2882, "lr": 3.568905675669895e-05, "epoch": 1.2707423580786026, "percentage": 42.36, "elapsed_time": "23:37:02", "remaining_time": "1 day, 8:08:20"} +{"current_steps": 2915, "total_steps": 6870, "loss": 0.2664, "lr": 3.5631607730305694e-05, "epoch": 1.2729257641921397, "percentage": 42.43, "elapsed_time": "23:39:28", "remaining_time": "1 day, 8:05:54"} +{"current_steps": 2920, "total_steps": 6870, "loss": 0.2798, "lr": 3.5574090085609696e-05, "epoch": 1.2751091703056767, "percentage": 42.5, "elapsed_time": "23:41:54", "remaining_time": "1 day, 8:03:28"} +{"current_steps": 2925, "total_steps": 6870, "loss": 0.2833, "lr": 3.551650419384013e-05, "epoch": 1.277292576419214, "percentage": 42.58, "elapsed_time": "23:44:20", "remaining_time": "1 day, 8:01:02"} +{"current_steps": 2930, "total_steps": 6870, "loss": 0.2738, "lr": 3.545885042666669e-05, "epoch": 1.279475982532751, "percentage": 42.65, "elapsed_time": "23:46:47", "remaining_time": "1 day, 7:58:36"} +{"current_steps": 2935, "total_steps": 6870, "loss": 0.2833, "lr": 3.5401129156197105e-05, "epoch": 1.2816593886462881, "percentage": 42.72, "elapsed_time": "23:49:13", "remaining_time": "1 day, 7:56:10"} +{"current_steps": 2940, "total_steps": 6870, "loss": 0.2821, "lr": 3.534334075497482e-05, "epoch": 1.2838427947598254, "percentage": 42.79, "elapsed_time": "23:51:39", "remaining_time": "1 day, 7:53:44"} +{"current_steps": 2945, "total_steps": 6870, "loss": 0.2821, "lr": 3.5285485595976534e-05, "epoch": 1.2860262008733625, "percentage": 42.87, "elapsed_time": "23:54:05", "remaining_time": "1 day, 7:51:19"} +{"current_steps": 2950, "total_steps": 6870, "loss": 0.2724, "lr": 3.5227564052609816e-05, "epoch": 1.2882096069868996, "percentage": 42.94, "elapsed_time": "23:56:31", "remaining_time": "1 day, 7:48:53"} +{"current_steps": 2955, "total_steps": 6870, "loss": 0.278, "lr": 3.5169576498710696e-05, "epoch": 1.2903930131004366, "percentage": 43.01, "elapsed_time": "23:58:58", "remaining_time": "1 day, 7:46:26"} +{"current_steps": 2960, "total_steps": 6870, "loss": 0.2679, "lr": 3.511152330854125e-05, "epoch": 1.2925764192139737, "percentage": 43.09, "elapsed_time": "1 day, 0:01:24", "remaining_time": "1 day, 7:44:00"} +{"current_steps": 2965, "total_steps": 6870, "loss": 0.2819, "lr": 3.505340485678717e-05, "epoch": 1.294759825327511, "percentage": 43.16, "elapsed_time": "1 day, 0:03:50", "remaining_time": "1 day, 7:41:34"} +{"current_steps": 2970, "total_steps": 6870, "loss": 0.2975, "lr": 3.499522151855537e-05, "epoch": 1.296943231441048, "percentage": 43.23, "elapsed_time": "1 day, 0:06:16", "remaining_time": "1 day, 7:39:08"} +{"current_steps": 2975, "total_steps": 6870, "loss": 0.2706, "lr": 3.493697366937154e-05, "epoch": 1.2991266375545851, "percentage": 43.3, "elapsed_time": "1 day, 0:08:42", "remaining_time": "1 day, 7:36:42"} +{"current_steps": 2980, "total_steps": 6870, "loss": 0.294, "lr": 3.487866168517775e-05, "epoch": 1.3013100436681222, "percentage": 43.38, "elapsed_time": "1 day, 0:11:08", "remaining_time": "1 day, 7:34:16"} +{"current_steps": 2985, "total_steps": 6870, "loss": 0.2818, "lr": 3.4820285942330003e-05, "epoch": 1.3034934497816595, "percentage": 43.45, "elapsed_time": "1 day, 0:13:34", "remaining_time": "1 day, 7:31:50"} +{"current_steps": 2990, "total_steps": 6870, "loss": 0.28, "lr": 3.476184681759581e-05, "epoch": 1.3056768558951966, "percentage": 43.52, "elapsed_time": "1 day, 0:16:00", "remaining_time": "1 day, 7:29:24"} +{"current_steps": 2995, "total_steps": 6870, "loss": 0.2783, "lr": 3.4703344688151776e-05, "epoch": 1.3078602620087336, "percentage": 43.6, "elapsed_time": "1 day, 0:18:26", "remaining_time": "1 day, 7:26:58"} +{"current_steps": 3000, "total_steps": 6870, "loss": 0.2819, "lr": 3.46447799315811e-05, "epoch": 1.3100436681222707, "percentage": 43.67, "elapsed_time": "1 day, 0:20:52", "remaining_time": "1 day, 7:24:32"} +{"current_steps": 3005, "total_steps": 6870, "loss": 0.2792, "lr": 3.458615292587125e-05, "epoch": 1.3122270742358078, "percentage": 43.74, "elapsed_time": "1 day, 0:23:18", "remaining_time": "1 day, 7:22:06"} +{"current_steps": 3010, "total_steps": 6870, "loss": 0.2859, "lr": 3.452746404941143e-05, "epoch": 1.314410480349345, "percentage": 43.81, "elapsed_time": "1 day, 0:25:45", "remaining_time": "1 day, 7:19:40"} +{"current_steps": 3015, "total_steps": 6870, "loss": 0.2746, "lr": 3.446871368099017e-05, "epoch": 1.3165938864628821, "percentage": 43.89, "elapsed_time": "1 day, 0:28:11", "remaining_time": "1 day, 7:17:13"} +{"current_steps": 3020, "total_steps": 6870, "loss": 0.2863, "lr": 3.440990219979289e-05, "epoch": 1.3187772925764192, "percentage": 43.96, "elapsed_time": "1 day, 0:30:37", "remaining_time": "1 day, 7:14:47"} +{"current_steps": 3025, "total_steps": 6870, "loss": 0.2702, "lr": 3.435102998539942e-05, "epoch": 1.3209606986899565, "percentage": 44.03, "elapsed_time": "1 day, 0:33:03", "remaining_time": "1 day, 7:12:21"} +{"current_steps": 3030, "total_steps": 6870, "loss": 0.2695, "lr": 3.4292097417781595e-05, "epoch": 1.3231441048034935, "percentage": 44.1, "elapsed_time": "1 day, 0:35:29", "remaining_time": "1 day, 7:09:55"} +{"current_steps": 3035, "total_steps": 6870, "loss": 0.267, "lr": 3.4233104877300775e-05, "epoch": 1.3253275109170306, "percentage": 44.18, "elapsed_time": "1 day, 0:37:55", "remaining_time": "1 day, 7:07:29"} +{"current_steps": 3040, "total_steps": 6870, "loss": 0.2662, "lr": 3.41740527447054e-05, "epoch": 1.3275109170305677, "percentage": 44.25, "elapsed_time": "1 day, 0:40:21", "remaining_time": "1 day, 7:05:03"} +{"current_steps": 3045, "total_steps": 6870, "loss": 0.2716, "lr": 3.411494140112852e-05, "epoch": 1.3296943231441047, "percentage": 44.32, "elapsed_time": "1 day, 0:42:47", "remaining_time": "1 day, 7:02:37"} +{"current_steps": 3050, "total_steps": 6870, "loss": 0.2879, "lr": 3.4055771228085347e-05, "epoch": 1.3318777292576418, "percentage": 44.4, "elapsed_time": "1 day, 0:45:13", "remaining_time": "1 day, 7:00:10"} +{"current_steps": 3055, "total_steps": 6870, "loss": 0.2725, "lr": 3.399654260747078e-05, "epoch": 1.334061135371179, "percentage": 44.47, "elapsed_time": "1 day, 0:47:39", "remaining_time": "1 day, 6:57:44"} +{"current_steps": 3060, "total_steps": 6870, "loss": 0.2877, "lr": 3.3937255921556974e-05, "epoch": 1.3362445414847162, "percentage": 44.54, "elapsed_time": "1 day, 0:50:05", "remaining_time": "1 day, 6:55:18"} +{"current_steps": 3065, "total_steps": 6870, "loss": 0.2793, "lr": 3.3877911552990826e-05, "epoch": 1.3384279475982532, "percentage": 44.61, "elapsed_time": "1 day, 0:52:31", "remaining_time": "1 day, 6:52:52"} +{"current_steps": 3070, "total_steps": 6870, "loss": 0.273, "lr": 3.3818509884791514e-05, "epoch": 1.3406113537117905, "percentage": 44.69, "elapsed_time": "1 day, 0:54:57", "remaining_time": "1 day, 6:50:26"} +{"current_steps": 3075, "total_steps": 6870, "loss": 0.275, "lr": 3.375905130034808e-05, "epoch": 1.3427947598253276, "percentage": 44.76, "elapsed_time": "1 day, 0:57:23", "remaining_time": "1 day, 6:48:00"} +{"current_steps": 3080, "total_steps": 6870, "loss": 0.2874, "lr": 3.369953618341688e-05, "epoch": 1.3449781659388647, "percentage": 44.83, "elapsed_time": "1 day, 0:59:50", "remaining_time": "1 day, 6:45:34"} +{"current_steps": 3085, "total_steps": 6870, "loss": 0.2756, "lr": 3.3639964918119135e-05, "epoch": 1.3471615720524017, "percentage": 44.91, "elapsed_time": "1 day, 1:02:16", "remaining_time": "1 day, 6:43:08"} +{"current_steps": 3090, "total_steps": 6870, "loss": 0.2795, "lr": 3.3580337888938515e-05, "epoch": 1.3493449781659388, "percentage": 44.98, "elapsed_time": "1 day, 1:04:42", "remaining_time": "1 day, 6:40:42"} +{"current_steps": 3095, "total_steps": 6870, "loss": 0.2821, "lr": 3.3520655480718515e-05, "epoch": 1.3515283842794759, "percentage": 45.05, "elapsed_time": "1 day, 1:07:08", "remaining_time": "1 day, 6:38:16"} +{"current_steps": 3100, "total_steps": 6870, "loss": 0.2799, "lr": 3.346091807866013e-05, "epoch": 1.3537117903930131, "percentage": 45.12, "elapsed_time": "1 day, 1:09:34", "remaining_time": "1 day, 6:35:50"} +{"current_steps": 3105, "total_steps": 6870, "loss": 0.2753, "lr": 3.3401126068319244e-05, "epoch": 1.3558951965065502, "percentage": 45.2, "elapsed_time": "1 day, 1:12:00", "remaining_time": "1 day, 6:33:24"} +{"current_steps": 3110, "total_steps": 6870, "loss": 0.2806, "lr": 3.3341279835604224e-05, "epoch": 1.3580786026200873, "percentage": 45.27, "elapsed_time": "1 day, 1:14:26", "remaining_time": "1 day, 6:30:58"} +{"current_steps": 3115, "total_steps": 6870, "loss": 0.2952, "lr": 3.328137976677339e-05, "epoch": 1.3602620087336246, "percentage": 45.34, "elapsed_time": "1 day, 1:16:52", "remaining_time": "1 day, 6:28:32"} +{"current_steps": 3120, "total_steps": 6870, "loss": 0.2883, "lr": 3.322142624843253e-05, "epoch": 1.3624454148471616, "percentage": 45.41, "elapsed_time": "1 day, 1:19:19", "remaining_time": "1 day, 6:26:06"} +{"current_steps": 3125, "total_steps": 6870, "loss": 0.2811, "lr": 3.3161419667532384e-05, "epoch": 1.3646288209606987, "percentage": 45.49, "elapsed_time": "1 day, 1:21:45", "remaining_time": "1 day, 6:23:40"} +{"current_steps": 3130, "total_steps": 6870, "loss": 0.2714, "lr": 3.3101360411366205e-05, "epoch": 1.3668122270742358, "percentage": 45.56, "elapsed_time": "1 day, 1:24:11", "remaining_time": "1 day, 6:21:14"} +{"current_steps": 3135, "total_steps": 6870, "loss": 0.2773, "lr": 3.304124886756719e-05, "epoch": 1.3689956331877728, "percentage": 45.63, "elapsed_time": "1 day, 1:26:37", "remaining_time": "1 day, 6:18:48"} +{"current_steps": 3140, "total_steps": 6870, "loss": 0.2716, "lr": 3.2981085424106036e-05, "epoch": 1.37117903930131, "percentage": 45.71, "elapsed_time": "1 day, 1:29:03", "remaining_time": "1 day, 6:16:22"} +{"current_steps": 3145, "total_steps": 6870, "loss": 0.2929, "lr": 3.2920870469288375e-05, "epoch": 1.3733624454148472, "percentage": 45.78, "elapsed_time": "1 day, 1:31:29", "remaining_time": "1 day, 6:13:56"} +{"current_steps": 3150, "total_steps": 6870, "loss": 0.2767, "lr": 3.2860604391752344e-05, "epoch": 1.3755458515283843, "percentage": 45.85, "elapsed_time": "1 day, 1:33:55", "remaining_time": "1 day, 6:11:29"} +{"current_steps": 3155, "total_steps": 6870, "loss": 0.2803, "lr": 3.2800287580466e-05, "epoch": 1.3777292576419213, "percentage": 45.92, "elapsed_time": "1 day, 1:36:21", "remaining_time": "1 day, 6:09:03"} +{"current_steps": 3160, "total_steps": 6870, "loss": 0.2635, "lr": 3.273992042472487e-05, "epoch": 1.3799126637554586, "percentage": 46.0, "elapsed_time": "1 day, 1:38:48", "remaining_time": "1 day, 6:06:37"} +{"current_steps": 3165, "total_steps": 6870, "loss": 0.2828, "lr": 3.2679503314149395e-05, "epoch": 1.3820960698689957, "percentage": 46.07, "elapsed_time": "1 day, 1:41:14", "remaining_time": "1 day, 6:04:11"} +{"current_steps": 3170, "total_steps": 6870, "loss": 0.2772, "lr": 3.261903663868243e-05, "epoch": 1.3842794759825328, "percentage": 46.14, "elapsed_time": "1 day, 1:43:40", "remaining_time": "1 day, 6:01:45"} +{"current_steps": 3175, "total_steps": 6870, "loss": 0.2709, "lr": 3.2558520788586764e-05, "epoch": 1.3864628820960698, "percentage": 46.22, "elapsed_time": "1 day, 1:46:06", "remaining_time": "1 day, 5:59:19"} +{"current_steps": 3180, "total_steps": 6870, "loss": 0.2649, "lr": 3.249795615444253e-05, "epoch": 1.388646288209607, "percentage": 46.29, "elapsed_time": "1 day, 1:48:32", "remaining_time": "1 day, 5:56:53"} +{"current_steps": 3185, "total_steps": 6870, "loss": 0.2767, "lr": 3.243734312714475e-05, "epoch": 1.3908296943231442, "percentage": 46.36, "elapsed_time": "1 day, 1:50:58", "remaining_time": "1 day, 5:54:27"} +{"current_steps": 3190, "total_steps": 6870, "loss": 0.2793, "lr": 3.2376682097900754e-05, "epoch": 1.3930131004366813, "percentage": 46.43, "elapsed_time": "1 day, 1:53:24", "remaining_time": "1 day, 5:52:01"} +{"current_steps": 3195, "total_steps": 6870, "loss": 0.2875, "lr": 3.2315973458227714e-05, "epoch": 1.3951965065502183, "percentage": 46.51, "elapsed_time": "1 day, 1:55:50", "remaining_time": "1 day, 5:49:35"} +{"current_steps": 3200, "total_steps": 6870, "loss": 0.2821, "lr": 3.225521759995007e-05, "epoch": 1.3973799126637554, "percentage": 46.58, "elapsed_time": "1 day, 1:58:16", "remaining_time": "1 day, 5:47:09"} +{"current_steps": 3205, "total_steps": 6870, "loss": 0.2742, "lr": 3.219441491519703e-05, "epoch": 1.3995633187772927, "percentage": 46.65, "elapsed_time": "1 day, 2:00:42", "remaining_time": "1 day, 5:44:43"} +{"current_steps": 3210, "total_steps": 6870, "loss": 0.2794, "lr": 3.2133565796400046e-05, "epoch": 1.4017467248908297, "percentage": 46.72, "elapsed_time": "1 day, 2:03:08", "remaining_time": "1 day, 5:42:16"} +{"current_steps": 3215, "total_steps": 6870, "loss": 0.2686, "lr": 3.2072670636290195e-05, "epoch": 1.4039301310043668, "percentage": 46.8, "elapsed_time": "1 day, 2:05:35", "remaining_time": "1 day, 5:39:50"} +{"current_steps": 3220, "total_steps": 6870, "loss": 0.2744, "lr": 3.20117298278958e-05, "epoch": 1.4061135371179039, "percentage": 46.87, "elapsed_time": "1 day, 2:08:01", "remaining_time": "1 day, 5:37:24"} +{"current_steps": 3225, "total_steps": 6870, "loss": 0.2904, "lr": 3.195074376453975e-05, "epoch": 1.408296943231441, "percentage": 46.94, "elapsed_time": "1 day, 2:10:27", "remaining_time": "1 day, 5:34:58"} +{"current_steps": 3230, "total_steps": 6870, "loss": 0.2633, "lr": 3.188971283983703e-05, "epoch": 1.4104803493449782, "percentage": 47.02, "elapsed_time": "1 day, 2:12:53", "remaining_time": "1 day, 5:32:32"} +{"current_steps": 3235, "total_steps": 6870, "loss": 0.2755, "lr": 3.1828637447692187e-05, "epoch": 1.4126637554585153, "percentage": 47.09, "elapsed_time": "1 day, 2:15:19", "remaining_time": "1 day, 5:30:06"} +{"current_steps": 3240, "total_steps": 6870, "loss": 0.2802, "lr": 3.176751798229673e-05, "epoch": 1.4148471615720524, "percentage": 47.16, "elapsed_time": "1 day, 2:17:45", "remaining_time": "1 day, 5:27:40"} +{"current_steps": 3245, "total_steps": 6870, "loss": 0.2585, "lr": 3.1706354838126687e-05, "epoch": 1.4170305676855894, "percentage": 47.23, "elapsed_time": "1 day, 2:20:11", "remaining_time": "1 day, 5:25:14"} +{"current_steps": 3250, "total_steps": 6870, "loss": 0.2758, "lr": 3.1645148409939915e-05, "epoch": 1.4192139737991267, "percentage": 47.31, "elapsed_time": "1 day, 2:22:37", "remaining_time": "1 day, 5:22:48"} +{"current_steps": 3255, "total_steps": 6870, "loss": 0.2857, "lr": 3.158389909277371e-05, "epoch": 1.4213973799126638, "percentage": 47.38, "elapsed_time": "1 day, 2:25:03", "remaining_time": "1 day, 5:20:22"} +{"current_steps": 3260, "total_steps": 6870, "loss": 0.2827, "lr": 3.152260728194213e-05, "epoch": 1.4235807860262009, "percentage": 47.45, "elapsed_time": "1 day, 2:27:29", "remaining_time": "1 day, 5:17:56"} +{"current_steps": 3265, "total_steps": 6870, "loss": 0.2791, "lr": 3.1461273373033516e-05, "epoch": 1.425764192139738, "percentage": 47.53, "elapsed_time": "1 day, 2:29:56", "remaining_time": "1 day, 5:15:30"} +{"current_steps": 3270, "total_steps": 6870, "loss": 0.2785, "lr": 3.1399897761907914e-05, "epoch": 1.427947598253275, "percentage": 47.6, "elapsed_time": "1 day, 2:32:21", "remaining_time": "1 day, 5:13:03"} +{"current_steps": 3275, "total_steps": 6870, "loss": 0.2723, "lr": 3.133848084469454e-05, "epoch": 1.4301310043668123, "percentage": 47.67, "elapsed_time": "1 day, 2:34:47", "remaining_time": "1 day, 5:10:37"} +{"current_steps": 3280, "total_steps": 6870, "loss": 0.2701, "lr": 3.1277023017789164e-05, "epoch": 1.4323144104803494, "percentage": 47.74, "elapsed_time": "1 day, 2:37:14", "remaining_time": "1 day, 5:08:11"} +{"current_steps": 3285, "total_steps": 6870, "loss": 0.2756, "lr": 3.121552467785164e-05, "epoch": 1.4344978165938864, "percentage": 47.82, "elapsed_time": "1 day, 2:39:40", "remaining_time": "1 day, 5:05:45"} +{"current_steps": 3290, "total_steps": 6870, "loss": 0.2634, "lr": 3.1153986221803284e-05, "epoch": 1.4366812227074237, "percentage": 47.89, "elapsed_time": "1 day, 2:42:06", "remaining_time": "1 day, 5:03:19"} +{"current_steps": 3295, "total_steps": 6870, "loss": 0.2749, "lr": 3.109240804682431e-05, "epoch": 1.4388646288209608, "percentage": 47.96, "elapsed_time": "1 day, 2:44:32", "remaining_time": "1 day, 5:00:53"} +{"current_steps": 3300, "total_steps": 6870, "loss": 0.2737, "lr": 3.103079055035131e-05, "epoch": 1.4410480349344978, "percentage": 48.03, "elapsed_time": "1 day, 2:46:58", "remaining_time": "1 day, 4:58:27"} +{"current_steps": 3305, "total_steps": 6870, "loss": 0.2716, "lr": 3.096913413007467e-05, "epoch": 1.443231441048035, "percentage": 48.11, "elapsed_time": "1 day, 2:49:24", "remaining_time": "1 day, 4:56:01"} +{"current_steps": 3310, "total_steps": 6870, "loss": 0.2711, "lr": 3.090743918393596e-05, "epoch": 1.445414847161572, "percentage": 48.18, "elapsed_time": "1 day, 2:51:50", "remaining_time": "1 day, 4:53:35"} +{"current_steps": 3315, "total_steps": 6870, "loss": 0.2755, "lr": 3.0845706110125435e-05, "epoch": 1.447598253275109, "percentage": 48.25, "elapsed_time": "1 day, 2:54:16", "remaining_time": "1 day, 4:51:09"} +{"current_steps": 3320, "total_steps": 6870, "loss": 0.2708, "lr": 3.078393530707942e-05, "epoch": 1.4497816593886463, "percentage": 48.33, "elapsed_time": "1 day, 2:56:42", "remaining_time": "1 day, 4:48:42"} +{"current_steps": 3325, "total_steps": 6870, "loss": 0.2802, "lr": 3.072212717347776e-05, "epoch": 1.4519650655021834, "percentage": 48.4, "elapsed_time": "1 day, 2:59:08", "remaining_time": "1 day, 4:46:16"} +{"current_steps": 3330, "total_steps": 6870, "loss": 0.2793, "lr": 3.0660282108241235e-05, "epoch": 1.4541484716157205, "percentage": 48.47, "elapsed_time": "1 day, 3:01:35", "remaining_time": "1 day, 4:43:50"} +{"current_steps": 3335, "total_steps": 6870, "loss": 0.2915, "lr": 3.059840051052898e-05, "epoch": 1.4563318777292578, "percentage": 48.54, "elapsed_time": "1 day, 3:04:01", "remaining_time": "1 day, 4:41:24"} +{"current_steps": 3340, "total_steps": 6870, "loss": 0.2667, "lr": 3.053648277973592e-05, "epoch": 1.4585152838427948, "percentage": 48.62, "elapsed_time": "1 day, 3:06:27", "remaining_time": "1 day, 4:38:58"} +{"current_steps": 3345, "total_steps": 6870, "loss": 0.2863, "lr": 3.04745293154902e-05, "epoch": 1.460698689956332, "percentage": 48.69, "elapsed_time": "1 day, 3:08:53", "remaining_time": "1 day, 4:36:32"} +{"current_steps": 3350, "total_steps": 6870, "loss": 0.2686, "lr": 3.0412540517650582e-05, "epoch": 1.462882096069869, "percentage": 48.76, "elapsed_time": "1 day, 3:11:19", "remaining_time": "1 day, 4:34:06"} +{"current_steps": 3355, "total_steps": 6870, "loss": 0.2861, "lr": 3.03505167863039e-05, "epoch": 1.465065502183406, "percentage": 48.84, "elapsed_time": "1 day, 3:13:45", "remaining_time": "1 day, 4:31:40"} +{"current_steps": 3360, "total_steps": 6870, "loss": 0.2804, "lr": 3.0288458521762414e-05, "epoch": 1.467248908296943, "percentage": 48.91, "elapsed_time": "1 day, 3:16:11", "remaining_time": "1 day, 4:29:14"} +{"current_steps": 3365, "total_steps": 6870, "loss": 0.2673, "lr": 3.02263661245613e-05, "epoch": 1.4694323144104804, "percentage": 48.98, "elapsed_time": "1 day, 3:18:37", "remaining_time": "1 day, 4:26:48"} +{"current_steps": 3370, "total_steps": 6870, "loss": 0.2805, "lr": 3.016423999545603e-05, "epoch": 1.4716157205240175, "percentage": 49.05, "elapsed_time": "1 day, 3:21:04", "remaining_time": "1 day, 4:24:22"} +{"current_steps": 3375, "total_steps": 6870, "loss": 0.2852, "lr": 3.0102080535419775e-05, "epoch": 1.4737991266375545, "percentage": 49.13, "elapsed_time": "1 day, 3:23:30", "remaining_time": "1 day, 4:21:56"} +{"current_steps": 3380, "total_steps": 6870, "loss": 0.2768, "lr": 3.0039888145640845e-05, "epoch": 1.4759825327510918, "percentage": 49.2, "elapsed_time": "1 day, 3:25:58", "remaining_time": "1 day, 4:19:32"} +{"current_steps": 3385, "total_steps": 6870, "loss": 0.2733, "lr": 2.9977663227520065e-05, "epoch": 1.4781659388646289, "percentage": 49.27, "elapsed_time": "1 day, 3:28:24", "remaining_time": "1 day, 4:17:06"} +{"current_steps": 3390, "total_steps": 6870, "loss": 0.2689, "lr": 2.991540618266822e-05, "epoch": 1.480349344978166, "percentage": 49.34, "elapsed_time": "1 day, 3:30:50", "remaining_time": "1 day, 4:14:39"} +{"current_steps": 3395, "total_steps": 6870, "loss": 0.259, "lr": 2.9853117412903443e-05, "epoch": 1.482532751091703, "percentage": 49.42, "elapsed_time": "1 day, 3:33:16", "remaining_time": "1 day, 4:12:13"} +{"current_steps": 3400, "total_steps": 6870, "loss": 0.2676, "lr": 2.979079732024862e-05, "epoch": 1.48471615720524, "percentage": 49.49, "elapsed_time": "1 day, 3:35:42", "remaining_time": "1 day, 4:09:47"} +{"current_steps": 3405, "total_steps": 6870, "loss": 0.274, "lr": 2.9728446306928813e-05, "epoch": 1.4868995633187774, "percentage": 49.56, "elapsed_time": "1 day, 3:38:08", "remaining_time": "1 day, 4:07:21"} +{"current_steps": 3410, "total_steps": 6870, "loss": 0.265, "lr": 2.966606477536863e-05, "epoch": 1.4890829694323144, "percentage": 49.64, "elapsed_time": "1 day, 3:40:34", "remaining_time": "1 day, 4:04:55"} +{"current_steps": 3415, "total_steps": 6870, "loss": 0.2745, "lr": 2.960365312818967e-05, "epoch": 1.4912663755458515, "percentage": 49.71, "elapsed_time": "1 day, 3:43:00", "remaining_time": "1 day, 4:02:29"} +{"current_steps": 3420, "total_steps": 6870, "loss": 0.2718, "lr": 2.9541211768207883e-05, "epoch": 1.4934497816593886, "percentage": 49.78, "elapsed_time": "1 day, 3:45:26", "remaining_time": "1 day, 4:00:03"} +{"current_steps": 3425, "total_steps": 6870, "loss": 0.2747, "lr": 2.9478741098431027e-05, "epoch": 1.4956331877729259, "percentage": 49.85, "elapsed_time": "1 day, 3:47:53", "remaining_time": "1 day, 3:57:37"} +{"current_steps": 3430, "total_steps": 6870, "loss": 0.3007, "lr": 2.9416241522055987e-05, "epoch": 1.497816593886463, "percentage": 49.93, "elapsed_time": "1 day, 3:50:19", "remaining_time": "1 day, 3:55:11"} +{"current_steps": 3435, "total_steps": 6870, "loss": 0.2797, "lr": 2.9353713442466246e-05, "epoch": 1.5, "percentage": 50.0, "elapsed_time": "1 day, 3:52:45", "remaining_time": "1 day, 3:52:45"} +{"current_steps": 3440, "total_steps": 6870, "loss": 0.2711, "lr": 2.929115726322924e-05, "epoch": 1.502183406113537, "percentage": 50.07, "elapsed_time": "1 day, 3:55:11", "remaining_time": "1 day, 3:50:19"} +{"current_steps": 3445, "total_steps": 6870, "loss": 0.2781, "lr": 2.9228573388093765e-05, "epoch": 1.5043668122270741, "percentage": 50.15, "elapsed_time": "1 day, 3:57:37", "remaining_time": "1 day, 3:47:53"} +{"current_steps": 3450, "total_steps": 6870, "loss": 0.2782, "lr": 2.916596222098738e-05, "epoch": 1.5065502183406112, "percentage": 50.22, "elapsed_time": "1 day, 4:00:03", "remaining_time": "1 day, 3:45:27"} +{"current_steps": 3455, "total_steps": 6870, "loss": 0.2723, "lr": 2.9103324166013777e-05, "epoch": 1.5087336244541485, "percentage": 50.29, "elapsed_time": "1 day, 4:02:29", "remaining_time": "1 day, 3:43:01"} +{"current_steps": 3460, "total_steps": 6870, "loss": 0.2734, "lr": 2.9040659627450195e-05, "epoch": 1.5109170305676856, "percentage": 50.36, "elapsed_time": "1 day, 4:04:55", "remaining_time": "1 day, 3:40:34"} +{"current_steps": 3465, "total_steps": 6870, "loss": 0.2695, "lr": 2.897796900974481e-05, "epoch": 1.5131004366812228, "percentage": 50.44, "elapsed_time": "1 day, 4:07:22", "remaining_time": "1 day, 3:38:08"} +{"current_steps": 3470, "total_steps": 6870, "loss": 0.2753, "lr": 2.8915252717514108e-05, "epoch": 1.51528384279476, "percentage": 50.51, "elapsed_time": "1 day, 4:09:48", "remaining_time": "1 day, 3:35:42"} +{"current_steps": 3475, "total_steps": 6870, "loss": 0.2805, "lr": 2.8852511155540286e-05, "epoch": 1.517467248908297, "percentage": 50.58, "elapsed_time": "1 day, 4:12:14", "remaining_time": "1 day, 3:33:16"} +{"current_steps": 3480, "total_steps": 6870, "loss": 0.2674, "lr": 2.8789744728768632e-05, "epoch": 1.519650655021834, "percentage": 50.66, "elapsed_time": "1 day, 4:14:40", "remaining_time": "1 day, 3:30:50"} +{"current_steps": 3485, "total_steps": 6870, "loss": 0.2781, "lr": 2.8726953842304917e-05, "epoch": 1.5218340611353711, "percentage": 50.73, "elapsed_time": "1 day, 4:17:06", "remaining_time": "1 day, 3:28:24"} +{"current_steps": 3490, "total_steps": 6870, "loss": 0.2828, "lr": 2.8664138901412784e-05, "epoch": 1.5240174672489082, "percentage": 50.8, "elapsed_time": "1 day, 4:19:32", "remaining_time": "1 day, 3:25:58"} +{"current_steps": 3495, "total_steps": 6870, "loss": 0.2704, "lr": 2.8601300311511135e-05, "epoch": 1.5262008733624453, "percentage": 50.87, "elapsed_time": "1 day, 4:21:58", "remaining_time": "1 day, 3:23:32"} +{"current_steps": 3500, "total_steps": 6870, "loss": 0.2709, "lr": 2.8538438478171493e-05, "epoch": 1.5283842794759825, "percentage": 50.95, "elapsed_time": "1 day, 4:24:24", "remaining_time": "1 day, 3:21:06"} +{"current_steps": 3505, "total_steps": 6870, "loss": 0.2645, "lr": 2.8475553807115392e-05, "epoch": 1.5305676855895196, "percentage": 51.02, "elapsed_time": "1 day, 4:26:50", "remaining_time": "1 day, 3:18:40"} +{"current_steps": 3510, "total_steps": 6870, "loss": 0.2716, "lr": 2.8412646704211787e-05, "epoch": 1.532751091703057, "percentage": 51.09, "elapsed_time": "1 day, 4:29:17", "remaining_time": "1 day, 3:16:14"} +{"current_steps": 3515, "total_steps": 6870, "loss": 0.2812, "lr": 2.8349717575474395e-05, "epoch": 1.534934497816594, "percentage": 51.16, "elapsed_time": "1 day, 4:31:43", "remaining_time": "1 day, 3:13:48"} +{"current_steps": 3520, "total_steps": 6870, "loss": 0.2852, "lr": 2.8286766827059103e-05, "epoch": 1.537117903930131, "percentage": 51.24, "elapsed_time": "1 day, 4:34:09", "remaining_time": "1 day, 3:11:21"} +{"current_steps": 3525, "total_steps": 6870, "loss": 0.2706, "lr": 2.8223794865261326e-05, "epoch": 1.539301310043668, "percentage": 51.31, "elapsed_time": "1 day, 4:36:35", "remaining_time": "1 day, 3:08:55"} +{"current_steps": 3530, "total_steps": 6870, "loss": 0.2655, "lr": 2.8160802096513396e-05, "epoch": 1.5414847161572052, "percentage": 51.38, "elapsed_time": "1 day, 4:39:01", "remaining_time": "1 day, 3:06:29"} +{"current_steps": 3535, "total_steps": 6870, "loss": 0.2799, "lr": 2.809778892738194e-05, "epoch": 1.5436681222707422, "percentage": 51.46, "elapsed_time": "1 day, 4:41:27", "remaining_time": "1 day, 3:04:03"} +{"current_steps": 3540, "total_steps": 6870, "loss": 0.2703, "lr": 2.8034755764565252e-05, "epoch": 1.5458515283842795, "percentage": 51.53, "elapsed_time": "1 day, 4:43:53", "remaining_time": "1 day, 3:01:37"} +{"current_steps": 3545, "total_steps": 6870, "loss": 0.2716, "lr": 2.797170301489066e-05, "epoch": 1.5480349344978166, "percentage": 51.6, "elapsed_time": "1 day, 4:46:19", "remaining_time": "1 day, 2:59:11"} +{"current_steps": 3550, "total_steps": 6870, "loss": 0.2617, "lr": 2.7908631085311933e-05, "epoch": 1.5502183406113537, "percentage": 51.67, "elapsed_time": "1 day, 4:48:45", "remaining_time": "1 day, 2:56:45"} +{"current_steps": 3555, "total_steps": 6870, "loss": 0.2638, "lr": 2.78455403829066e-05, "epoch": 1.552401746724891, "percentage": 51.75, "elapsed_time": "1 day, 4:51:11", "remaining_time": "1 day, 2:54:19"} +{"current_steps": 3560, "total_steps": 6870, "loss": 0.2678, "lr": 2.7782431314873377e-05, "epoch": 1.554585152838428, "percentage": 51.82, "elapsed_time": "1 day, 4:53:38", "remaining_time": "1 day, 2:51:53"} +{"current_steps": 3565, "total_steps": 6870, "loss": 0.2714, "lr": 2.7719304288529503e-05, "epoch": 1.556768558951965, "percentage": 51.89, "elapsed_time": "1 day, 4:56:04", "remaining_time": "1 day, 2:49:27"} +{"current_steps": 3570, "total_steps": 6870, "loss": 0.2754, "lr": 2.7656159711308127e-05, "epoch": 1.5589519650655022, "percentage": 51.97, "elapsed_time": "1 day, 4:58:30", "remaining_time": "1 day, 2:47:01"} +{"current_steps": 3575, "total_steps": 6870, "loss": 0.2674, "lr": 2.7592997990755665e-05, "epoch": 1.5611353711790392, "percentage": 52.04, "elapsed_time": "1 day, 5:00:56", "remaining_time": "1 day, 2:44:35"} +{"current_steps": 3580, "total_steps": 6870, "loss": 0.2689, "lr": 2.75298195345292e-05, "epoch": 1.5633187772925763, "percentage": 52.11, "elapsed_time": "1 day, 5:03:22", "remaining_time": "1 day, 2:42:09"} +{"current_steps": 3585, "total_steps": 6870, "loss": 0.2624, "lr": 2.7466624750393804e-05, "epoch": 1.5655021834061136, "percentage": 52.18, "elapsed_time": "1 day, 5:05:48", "remaining_time": "1 day, 2:39:43"} +{"current_steps": 3590, "total_steps": 6870, "loss": 0.2704, "lr": 2.740341404621996e-05, "epoch": 1.5676855895196506, "percentage": 52.26, "elapsed_time": "1 day, 5:08:14", "remaining_time": "1 day, 2:37:17"} +{"current_steps": 3595, "total_steps": 6870, "loss": 0.2596, "lr": 2.7340187829980884e-05, "epoch": 1.569868995633188, "percentage": 52.33, "elapsed_time": "1 day, 5:10:41", "remaining_time": "1 day, 2:34:51"} +{"current_steps": 3600, "total_steps": 6870, "loss": 0.2753, "lr": 2.7276946509749905e-05, "epoch": 1.572052401746725, "percentage": 52.4, "elapsed_time": "1 day, 5:13:07", "remaining_time": "1 day, 2:32:25"} +{"current_steps": 3605, "total_steps": 6870, "loss": 0.2813, "lr": 2.7213690493697852e-05, "epoch": 1.574235807860262, "percentage": 52.47, "elapsed_time": "1 day, 5:15:33", "remaining_time": "1 day, 2:29:58"} +{"current_steps": 3610, "total_steps": 6870, "loss": 0.2654, "lr": 2.7150420190090397e-05, "epoch": 1.5764192139737991, "percentage": 52.55, "elapsed_time": "1 day, 5:17:59", "remaining_time": "1 day, 2:27:32"} +{"current_steps": 3615, "total_steps": 6870, "loss": 0.269, "lr": 2.7087136007285417e-05, "epoch": 1.5786026200873362, "percentage": 52.62, "elapsed_time": "1 day, 5:20:25", "remaining_time": "1 day, 2:25:06"} +{"current_steps": 3620, "total_steps": 6870, "loss": 0.2603, "lr": 2.7023838353730386e-05, "epoch": 1.5807860262008733, "percentage": 52.69, "elapsed_time": "1 day, 5:22:51", "remaining_time": "1 day, 2:22:40"} +{"current_steps": 3625, "total_steps": 6870, "loss": 0.2615, "lr": 2.6960527637959704e-05, "epoch": 1.5829694323144103, "percentage": 52.77, "elapsed_time": "1 day, 5:25:18", "remaining_time": "1 day, 2:20:14"} +{"current_steps": 3630, "total_steps": 6870, "loss": 0.2745, "lr": 2.689720426859208e-05, "epoch": 1.5851528384279476, "percentage": 52.84, "elapsed_time": "1 day, 5:27:44", "remaining_time": "1 day, 2:17:48"} +{"current_steps": 3635, "total_steps": 6870, "loss": 0.2754, "lr": 2.68338686543279e-05, "epoch": 1.5873362445414847, "percentage": 52.91, "elapsed_time": "1 day, 5:30:10", "remaining_time": "1 day, 2:15:22"} +{"current_steps": 3640, "total_steps": 6870, "loss": 0.2687, "lr": 2.6770521203946582e-05, "epoch": 1.589519650655022, "percentage": 52.98, "elapsed_time": "1 day, 5:32:36", "remaining_time": "1 day, 2:12:56"} +{"current_steps": 3645, "total_steps": 6870, "loss": 0.2708, "lr": 2.670716232630392e-05, "epoch": 1.591703056768559, "percentage": 53.06, "elapsed_time": "1 day, 5:35:02", "remaining_time": "1 day, 2:10:30"} +{"current_steps": 3650, "total_steps": 6870, "loss": 0.2691, "lr": 2.6643792430329473e-05, "epoch": 1.5938864628820961, "percentage": 53.13, "elapsed_time": "1 day, 5:37:28", "remaining_time": "1 day, 2:08:04"} +{"current_steps": 3655, "total_steps": 6870, "loss": 0.2895, "lr": 2.6580411925023916e-05, "epoch": 1.5960698689956332, "percentage": 53.2, "elapsed_time": "1 day, 5:39:55", "remaining_time": "1 day, 2:05:38"} +{"current_steps": 3660, "total_steps": 6870, "loss": 0.2768, "lr": 2.6517021219456402e-05, "epoch": 1.5982532751091703, "percentage": 53.28, "elapsed_time": "1 day, 5:42:21", "remaining_time": "1 day, 2:03:12"} +{"current_steps": 3665, "total_steps": 6870, "loss": 0.2772, "lr": 2.6453620722761896e-05, "epoch": 1.6004366812227073, "percentage": 53.35, "elapsed_time": "1 day, 5:44:47", "remaining_time": "1 day, 2:00:46"} +{"current_steps": 3670, "total_steps": 6870, "loss": 0.2772, "lr": 2.6390210844138593e-05, "epoch": 1.6026200873362444, "percentage": 53.42, "elapsed_time": "1 day, 5:47:13", "remaining_time": "1 day, 1:58:20"} +{"current_steps": 3675, "total_steps": 6870, "loss": 0.2746, "lr": 2.6326791992845202e-05, "epoch": 1.6048034934497817, "percentage": 53.49, "elapsed_time": "1 day, 5:49:39", "remaining_time": "1 day, 1:55:54"} +{"current_steps": 3680, "total_steps": 6870, "loss": 0.2627, "lr": 2.626336457819837e-05, "epoch": 1.6069868995633187, "percentage": 53.57, "elapsed_time": "1 day, 5:52:05", "remaining_time": "1 day, 1:53:28"} +{"current_steps": 3685, "total_steps": 6870, "loss": 0.2896, "lr": 2.619992900957e-05, "epoch": 1.609170305676856, "percentage": 53.64, "elapsed_time": "1 day, 5:54:32", "remaining_time": "1 day, 1:51:02"} +{"current_steps": 3690, "total_steps": 6870, "loss": 0.2832, "lr": 2.613648569638464e-05, "epoch": 1.611353711790393, "percentage": 53.71, "elapsed_time": "1 day, 5:56:58", "remaining_time": "1 day, 1:48:36"} +{"current_steps": 3695, "total_steps": 6870, "loss": 0.2693, "lr": 2.6073035048116796e-05, "epoch": 1.6135371179039302, "percentage": 53.78, "elapsed_time": "1 day, 5:59:24", "remaining_time": "1 day, 1:46:10"} +{"current_steps": 3700, "total_steps": 6870, "loss": 0.2759, "lr": 2.6009577474288333e-05, "epoch": 1.6157205240174672, "percentage": 53.86, "elapsed_time": "1 day, 6:01:50", "remaining_time": "1 day, 1:43:44"} +{"current_steps": 3705, "total_steps": 6870, "loss": 0.2642, "lr": 2.5946113384465813e-05, "epoch": 1.6179039301310043, "percentage": 53.93, "elapsed_time": "1 day, 6:04:16", "remaining_time": "1 day, 1:41:18"} +{"current_steps": 3710, "total_steps": 6870, "loss": 0.2706, "lr": 2.5882643188257856e-05, "epoch": 1.6200873362445414, "percentage": 54.0, "elapsed_time": "1 day, 6:06:42", "remaining_time": "1 day, 1:38:52"} +{"current_steps": 3715, "total_steps": 6870, "loss": 0.2724, "lr": 2.5819167295312485e-05, "epoch": 1.6222707423580784, "percentage": 54.08, "elapsed_time": "1 day, 6:09:08", "remaining_time": "1 day, 1:36:26"} +{"current_steps": 3720, "total_steps": 6870, "loss": 0.2707, "lr": 2.57556861153145e-05, "epoch": 1.6244541484716157, "percentage": 54.15, "elapsed_time": "1 day, 6:11:35", "remaining_time": "1 day, 1:34:00"} +{"current_steps": 3725, "total_steps": 6870, "loss": 0.2725, "lr": 2.569220005798281e-05, "epoch": 1.6266375545851528, "percentage": 54.22, "elapsed_time": "1 day, 6:14:01", "remaining_time": "1 day, 1:31:34"} +{"current_steps": 3730, "total_steps": 6870, "loss": 0.2713, "lr": 2.5628709533067824e-05, "epoch": 1.62882096069869, "percentage": 54.29, "elapsed_time": "1 day, 6:16:27", "remaining_time": "1 day, 1:29:08"} +{"current_steps": 3735, "total_steps": 6870, "loss": 0.2606, "lr": 2.556521495034877e-05, "epoch": 1.6310043668122272, "percentage": 54.37, "elapsed_time": "1 day, 6:18:53", "remaining_time": "1 day, 1:26:42"} +{"current_steps": 3740, "total_steps": 6870, "loss": 0.2575, "lr": 2.550171671963108e-05, "epoch": 1.6331877729257642, "percentage": 54.44, "elapsed_time": "1 day, 6:21:19", "remaining_time": "1 day, 1:24:16"} +{"current_steps": 3745, "total_steps": 6870, "loss": 0.2699, "lr": 2.543821525074371e-05, "epoch": 1.6353711790393013, "percentage": 54.51, "elapsed_time": "1 day, 6:23:46", "remaining_time": "1 day, 1:21:50"} +{"current_steps": 3750, "total_steps": 6870, "loss": 0.2604, "lr": 2.537471095353653e-05, "epoch": 1.6375545851528384, "percentage": 54.59, "elapsed_time": "1 day, 6:26:12", "remaining_time": "1 day, 1:19:24"} +{"current_steps": 3755, "total_steps": 6870, "loss": 0.2715, "lr": 2.5311204237877657e-05, "epoch": 1.6397379912663754, "percentage": 54.66, "elapsed_time": "1 day, 6:28:38", "remaining_time": "1 day, 1:16:58"} +{"current_steps": 3760, "total_steps": 6870, "loss": 0.2569, "lr": 2.524769551365083e-05, "epoch": 1.6419213973799127, "percentage": 54.73, "elapsed_time": "1 day, 6:31:04", "remaining_time": "1 day, 1:14:31"} +{"current_steps": 3765, "total_steps": 6870, "loss": 0.2805, "lr": 2.5184185190752752e-05, "epoch": 1.6441048034934498, "percentage": 54.8, "elapsed_time": "1 day, 6:33:30", "remaining_time": "1 day, 1:12:05"} +{"current_steps": 3770, "total_steps": 6870, "loss": 0.2723, "lr": 2.512067367909041e-05, "epoch": 1.6462882096069869, "percentage": 54.88, "elapsed_time": "1 day, 6:35:56", "remaining_time": "1 day, 1:09:39"} +{"current_steps": 3775, "total_steps": 6870, "loss": 0.2692, "lr": 2.505716138857851e-05, "epoch": 1.6484716157205241, "percentage": 54.95, "elapsed_time": "1 day, 6:38:22", "remaining_time": "1 day, 1:07:13"} +{"current_steps": 3780, "total_steps": 6870, "loss": 0.2736, "lr": 2.499364872913676e-05, "epoch": 1.6506550218340612, "percentage": 55.02, "elapsed_time": "1 day, 6:40:49", "remaining_time": "1 day, 1:04:47"} +{"current_steps": 3785, "total_steps": 6870, "loss": 0.2711, "lr": 2.4930136110687246e-05, "epoch": 1.6528384279475983, "percentage": 55.09, "elapsed_time": "1 day, 6:43:15", "remaining_time": "1 day, 1:02:21"} +{"current_steps": 3790, "total_steps": 6870, "loss": 0.2865, "lr": 2.486662394315181e-05, "epoch": 1.6550218340611353, "percentage": 55.17, "elapsed_time": "1 day, 6:45:41", "remaining_time": "1 day, 0:59:55"} +{"current_steps": 3795, "total_steps": 6870, "loss": 0.2731, "lr": 2.4803112636449355e-05, "epoch": 1.6572052401746724, "percentage": 55.24, "elapsed_time": "1 day, 6:48:07", "remaining_time": "1 day, 0:57:29"} +{"current_steps": 3800, "total_steps": 6870, "loss": 0.2789, "lr": 2.473960260049325e-05, "epoch": 1.6593886462882095, "percentage": 55.31, "elapsed_time": "1 day, 6:50:33", "remaining_time": "1 day, 0:55:03"} +{"current_steps": 3805, "total_steps": 6870, "loss": 0.2933, "lr": 2.4676094245188658e-05, "epoch": 1.6615720524017468, "percentage": 55.39, "elapsed_time": "1 day, 6:53:00", "remaining_time": "1 day, 0:52:37"} +{"current_steps": 3810, "total_steps": 6870, "loss": 0.2685, "lr": 2.461258798042989e-05, "epoch": 1.6637554585152838, "percentage": 55.46, "elapsed_time": "1 day, 6:55:26", "remaining_time": "1 day, 0:50:11"} +{"current_steps": 3815, "total_steps": 6870, "loss": 0.2552, "lr": 2.454908421609777e-05, "epoch": 1.6659388646288211, "percentage": 55.53, "elapsed_time": "1 day, 6:57:52", "remaining_time": "1 day, 0:47:45"} +{"current_steps": 3820, "total_steps": 6870, "loss": 0.2786, "lr": 2.4485583362056974e-05, "epoch": 1.6681222707423582, "percentage": 55.6, "elapsed_time": "1 day, 7:00:18", "remaining_time": "1 day, 0:45:19"} +{"current_steps": 3825, "total_steps": 6870, "loss": 0.2662, "lr": 2.4422085828153403e-05, "epoch": 1.6703056768558953, "percentage": 55.68, "elapsed_time": "1 day, 7:02:44", "remaining_time": "1 day, 0:42:53"} +{"current_steps": 3830, "total_steps": 6870, "loss": 0.269, "lr": 2.435859202421153e-05, "epoch": 1.6724890829694323, "percentage": 55.75, "elapsed_time": "1 day, 7:05:11", "remaining_time": "1 day, 0:40:27"} +{"current_steps": 3835, "total_steps": 6870, "loss": 0.2597, "lr": 2.429510236003175e-05, "epoch": 1.6746724890829694, "percentage": 55.82, "elapsed_time": "1 day, 7:07:37", "remaining_time": "1 day, 0:38:01"} +{"current_steps": 3840, "total_steps": 6870, "loss": 0.2709, "lr": 2.4231617245387745e-05, "epoch": 1.6768558951965065, "percentage": 55.9, "elapsed_time": "1 day, 7:10:03", "remaining_time": "1 day, 0:35:35"} +{"current_steps": 3845, "total_steps": 6870, "loss": 0.2712, "lr": 2.416813709002382e-05, "epoch": 1.6790393013100435, "percentage": 55.97, "elapsed_time": "1 day, 7:12:29", "remaining_time": "1 day, 0:33:09"} +{"current_steps": 3850, "total_steps": 6870, "loss": 0.286, "lr": 2.4104662303652288e-05, "epoch": 1.6812227074235808, "percentage": 56.04, "elapsed_time": "1 day, 7:14:55", "remaining_time": "1 day, 0:30:43"} +{"current_steps": 3855, "total_steps": 6870, "loss": 0.2679, "lr": 2.4041193295950788e-05, "epoch": 1.6834061135371179, "percentage": 56.11, "elapsed_time": "1 day, 7:17:21", "remaining_time": "1 day, 0:28:17"} +{"current_steps": 3860, "total_steps": 6870, "loss": 0.2764, "lr": 2.39777304765597e-05, "epoch": 1.6855895196506552, "percentage": 56.19, "elapsed_time": "1 day, 7:19:48", "remaining_time": "1 day, 0:25:51"} +{"current_steps": 3865, "total_steps": 6870, "loss": 0.2773, "lr": 2.391427425507943e-05, "epoch": 1.6877729257641922, "percentage": 56.26, "elapsed_time": "1 day, 7:22:14", "remaining_time": "1 day, 0:23:25"} +{"current_steps": 3870, "total_steps": 6870, "loss": 0.2655, "lr": 2.385082504106781e-05, "epoch": 1.6899563318777293, "percentage": 56.33, "elapsed_time": "1 day, 7:24:40", "remaining_time": "1 day, 0:20:59"} +{"current_steps": 3875, "total_steps": 6870, "loss": 0.2649, "lr": 2.378738324403744e-05, "epoch": 1.6921397379912664, "percentage": 56.4, "elapsed_time": "1 day, 7:27:06", "remaining_time": "1 day, 0:18:33"} +{"current_steps": 3880, "total_steps": 6870, "loss": 0.2668, "lr": 2.3723949273453073e-05, "epoch": 1.6943231441048034, "percentage": 56.48, "elapsed_time": "1 day, 7:29:32", "remaining_time": "1 day, 0:16:07"} +{"current_steps": 3885, "total_steps": 6870, "loss": 0.2671, "lr": 2.3660523538728904e-05, "epoch": 1.6965065502183405, "percentage": 56.55, "elapsed_time": "1 day, 7:31:59", "remaining_time": "1 day, 0:13:41"} +{"current_steps": 3890, "total_steps": 6870, "loss": 0.2603, "lr": 2.3597106449226015e-05, "epoch": 1.6986899563318776, "percentage": 56.62, "elapsed_time": "1 day, 7:34:25", "remaining_time": "1 day, 0:11:15"} +{"current_steps": 3895, "total_steps": 6870, "loss": 0.2532, "lr": 2.3533698414249668e-05, "epoch": 1.7008733624454149, "percentage": 56.7, "elapsed_time": "1 day, 7:36:51", "remaining_time": "1 day, 0:08:49"} +{"current_steps": 3900, "total_steps": 6870, "loss": 0.2607, "lr": 2.3470299843046693e-05, "epoch": 1.703056768558952, "percentage": 56.77, "elapsed_time": "1 day, 7:39:17", "remaining_time": "1 day, 0:06:23"} +{"current_steps": 3905, "total_steps": 6870, "loss": 0.2675, "lr": 2.340691114480284e-05, "epoch": 1.7052401746724892, "percentage": 56.84, "elapsed_time": "1 day, 7:41:43", "remaining_time": "1 day, 0:03:57"} +{"current_steps": 3910, "total_steps": 6870, "loss": 0.2727, "lr": 2.334353272864013e-05, "epoch": 1.7074235807860263, "percentage": 56.91, "elapsed_time": "1 day, 7:44:09", "remaining_time": "1 day, 0:01:31"} +{"current_steps": 3915, "total_steps": 6870, "loss": 0.2663, "lr": 2.3280165003614227e-05, "epoch": 1.7096069868995634, "percentage": 56.99, "elapsed_time": "1 day, 7:46:36", "remaining_time": "23:59:04"} +{"current_steps": 3920, "total_steps": 6870, "loss": 0.2705, "lr": 2.3216808378711788e-05, "epoch": 1.7117903930131004, "percentage": 57.06, "elapsed_time": "1 day, 7:49:02", "remaining_time": "23:56:38"} +{"current_steps": 3925, "total_steps": 6870, "loss": 0.242, "lr": 2.3153463262847835e-05, "epoch": 1.7139737991266375, "percentage": 57.13, "elapsed_time": "1 day, 7:51:28", "remaining_time": "23:54:12"} +{"current_steps": 3930, "total_steps": 6870, "loss": 0.2681, "lr": 2.3090130064863113e-05, "epoch": 1.7161572052401746, "percentage": 57.21, "elapsed_time": "1 day, 7:53:54", "remaining_time": "23:51:46"} +{"current_steps": 3935, "total_steps": 6870, "loss": 0.2746, "lr": 2.3026809193521423e-05, "epoch": 1.7183406113537116, "percentage": 57.28, "elapsed_time": "1 day, 7:56:20", "remaining_time": "23:49:20"} +{"current_steps": 3940, "total_steps": 6870, "loss": 0.2682, "lr": 2.296350105750703e-05, "epoch": 1.720524017467249, "percentage": 57.35, "elapsed_time": "1 day, 7:58:47", "remaining_time": "23:46:54"} +{"current_steps": 3945, "total_steps": 6870, "loss": 0.26, "lr": 2.2900206065421997e-05, "epoch": 1.722707423580786, "percentage": 57.42, "elapsed_time": "1 day, 8:01:13", "remaining_time": "23:44:28"} +{"current_steps": 3950, "total_steps": 6870, "loss": 0.2776, "lr": 2.2836924625783554e-05, "epoch": 1.7248908296943233, "percentage": 57.5, "elapsed_time": "1 day, 8:03:39", "remaining_time": "23:42:02"} +{"current_steps": 3955, "total_steps": 6870, "loss": 0.281, "lr": 2.2773657147021464e-05, "epoch": 1.7270742358078603, "percentage": 57.57, "elapsed_time": "1 day, 8:06:05", "remaining_time": "23:39:36"} +{"current_steps": 3960, "total_steps": 6870, "loss": 0.2597, "lr": 2.271040403747537e-05, "epoch": 1.7292576419213974, "percentage": 57.64, "elapsed_time": "1 day, 8:08:31", "remaining_time": "23:37:10"} +{"current_steps": 3965, "total_steps": 6870, "loss": 0.2697, "lr": 2.2647165705392186e-05, "epoch": 1.7314410480349345, "percentage": 57.71, "elapsed_time": "1 day, 8:10:57", "remaining_time": "23:34:44"} +{"current_steps": 3970, "total_steps": 6870, "loss": 0.2716, "lr": 2.2583942558923456e-05, "epoch": 1.7336244541484715, "percentage": 57.79, "elapsed_time": "1 day, 8:13:24", "remaining_time": "23:32:18"} +{"current_steps": 3975, "total_steps": 6870, "loss": 0.274, "lr": 2.2520735006122688e-05, "epoch": 1.7358078602620086, "percentage": 57.86, "elapsed_time": "1 day, 8:15:50", "remaining_time": "23:29:52"} +{"current_steps": 3980, "total_steps": 6870, "loss": 0.2587, "lr": 2.2457543454942786e-05, "epoch": 1.737991266375546, "percentage": 57.93, "elapsed_time": "1 day, 8:18:16", "remaining_time": "23:27:26"} +{"current_steps": 3985, "total_steps": 6870, "loss": 0.2619, "lr": 2.2394368313233332e-05, "epoch": 1.740174672489083, "percentage": 58.01, "elapsed_time": "1 day, 8:20:42", "remaining_time": "23:25:00"} +{"current_steps": 3990, "total_steps": 6870, "loss": 0.2602, "lr": 2.2331209988738024e-05, "epoch": 1.74235807860262, "percentage": 58.08, "elapsed_time": "1 day, 8:23:08", "remaining_time": "23:22:34"} +{"current_steps": 3995, "total_steps": 6870, "loss": 0.262, "lr": 2.226806888909202e-05, "epoch": 1.7445414847161573, "percentage": 58.15, "elapsed_time": "1 day, 8:25:34", "remaining_time": "23:20:08"} +{"current_steps": 4000, "total_steps": 6870, "loss": 0.2636, "lr": 2.2204945421819292e-05, "epoch": 1.7467248908296944, "percentage": 58.22, "elapsed_time": "1 day, 8:28:01", "remaining_time": "23:17:42"} +{"current_steps": 4005, "total_steps": 6870, "loss": 0.2747, "lr": 2.2141839994330027e-05, "epoch": 1.7489082969432315, "percentage": 58.3, "elapsed_time": "1 day, 8:30:27", "remaining_time": "23:15:16"} +{"current_steps": 4010, "total_steps": 6870, "loss": 0.2706, "lr": 2.207875301391796e-05, "epoch": 1.7510917030567685, "percentage": 58.37, "elapsed_time": "1 day, 8:32:53", "remaining_time": "23:12:50"} +{"current_steps": 4015, "total_steps": 6870, "loss": 0.2544, "lr": 2.2015684887757776e-05, "epoch": 1.7532751091703056, "percentage": 58.44, "elapsed_time": "1 day, 8:35:19", "remaining_time": "23:10:24"} +{"current_steps": 4020, "total_steps": 6870, "loss": 0.2759, "lr": 2.195263602290248e-05, "epoch": 1.7554585152838427, "percentage": 58.52, "elapsed_time": "1 day, 8:37:45", "remaining_time": "23:07:57"} +{"current_steps": 4025, "total_steps": 6870, "loss": 0.2468, "lr": 2.188960682628074e-05, "epoch": 1.75764192139738, "percentage": 58.59, "elapsed_time": "1 day, 8:40:11", "remaining_time": "23:05:31"} +{"current_steps": 4030, "total_steps": 6870, "loss": 0.2699, "lr": 2.1826597704694306e-05, "epoch": 1.759825327510917, "percentage": 58.66, "elapsed_time": "1 day, 8:42:37", "remaining_time": "23:03:05"} +{"current_steps": 4035, "total_steps": 6870, "loss": 0.27, "lr": 2.176360906481533e-05, "epoch": 1.762008733624454, "percentage": 58.73, "elapsed_time": "1 day, 8:45:03", "remaining_time": "23:00:39"} +{"current_steps": 4040, "total_steps": 6870, "loss": 0.2604, "lr": 2.1700641313183807e-05, "epoch": 1.7641921397379914, "percentage": 58.81, "elapsed_time": "1 day, 8:47:30", "remaining_time": "22:58:13"} +{"current_steps": 4045, "total_steps": 6870, "loss": 0.2647, "lr": 2.163769485620489e-05, "epoch": 1.7663755458515285, "percentage": 58.88, "elapsed_time": "1 day, 8:49:56", "remaining_time": "22:55:47"} +{"current_steps": 4050, "total_steps": 6870, "loss": 0.2717, "lr": 2.1574770100146295e-05, "epoch": 1.7685589519650655, "percentage": 58.95, "elapsed_time": "1 day, 8:52:22", "remaining_time": "22:53:21"} +{"current_steps": 4055, "total_steps": 6870, "loss": 0.2599, "lr": 2.1511867451135693e-05, "epoch": 1.7707423580786026, "percentage": 59.02, "elapsed_time": "1 day, 8:54:48", "remaining_time": "22:50:55"} +{"current_steps": 4060, "total_steps": 6870, "loss": 0.2686, "lr": 2.1448987315158048e-05, "epoch": 1.7729257641921397, "percentage": 59.1, "elapsed_time": "1 day, 8:57:14", "remaining_time": "22:48:29"} +{"current_steps": 4065, "total_steps": 6870, "loss": 0.2627, "lr": 2.1386130098053036e-05, "epoch": 1.7751091703056767, "percentage": 59.17, "elapsed_time": "1 day, 8:59:40", "remaining_time": "22:46:03"} +{"current_steps": 4070, "total_steps": 6870, "loss": 0.2708, "lr": 2.1323296205512418e-05, "epoch": 1.777292576419214, "percentage": 59.24, "elapsed_time": "1 day, 9:02:06", "remaining_time": "22:43:36"} +{"current_steps": 4075, "total_steps": 6870, "loss": 0.2581, "lr": 2.1260486043077388e-05, "epoch": 1.779475982532751, "percentage": 59.32, "elapsed_time": "1 day, 9:04:33", "remaining_time": "22:41:10"} +{"current_steps": 4080, "total_steps": 6870, "loss": 0.266, "lr": 2.1197700016136014e-05, "epoch": 1.7816593886462884, "percentage": 59.39, "elapsed_time": "1 day, 9:06:59", "remaining_time": "22:38:44"} +{"current_steps": 4085, "total_steps": 6870, "loss": 0.2571, "lr": 2.113493852992056e-05, "epoch": 1.7838427947598254, "percentage": 59.46, "elapsed_time": "1 day, 9:09:25", "remaining_time": "22:36:18"} +{"current_steps": 4090, "total_steps": 6870, "loss": 0.2733, "lr": 2.1072201989504913e-05, "epoch": 1.7860262008733625, "percentage": 59.53, "elapsed_time": "1 day, 9:11:51", "remaining_time": "22:33:52"} +{"current_steps": 4095, "total_steps": 6870, "loss": 0.269, "lr": 2.1009490799801965e-05, "epoch": 1.7882096069868996, "percentage": 59.61, "elapsed_time": "1 day, 9:14:17", "remaining_time": "22:31:26"} +{"current_steps": 4100, "total_steps": 6870, "loss": 0.2745, "lr": 2.0946805365560966e-05, "epoch": 1.7903930131004366, "percentage": 59.68, "elapsed_time": "1 day, 9:16:43", "remaining_time": "22:29:00"} +{"current_steps": 4105, "total_steps": 6870, "loss": 0.2659, "lr": 2.0884146091364956e-05, "epoch": 1.7925764192139737, "percentage": 59.75, "elapsed_time": "1 day, 9:19:09", "remaining_time": "22:26:34"} +{"current_steps": 4110, "total_steps": 6870, "loss": 0.2645, "lr": 2.0821513381628132e-05, "epoch": 1.7947598253275108, "percentage": 59.83, "elapsed_time": "1 day, 9:21:36", "remaining_time": "22:24:08"} +{"current_steps": 4115, "total_steps": 6870, "loss": 0.2563, "lr": 2.0758907640593228e-05, "epoch": 1.796943231441048, "percentage": 59.9, "elapsed_time": "1 day, 9:24:02", "remaining_time": "22:21:42"} +{"current_steps": 4120, "total_steps": 6870, "loss": 0.2721, "lr": 2.0696329272328932e-05, "epoch": 1.7991266375545851, "percentage": 59.97, "elapsed_time": "1 day, 9:26:28", "remaining_time": "22:19:16"} +{"current_steps": 4125, "total_steps": 6870, "loss": 0.2616, "lr": 2.063377868072724e-05, "epoch": 1.8013100436681224, "percentage": 60.04, "elapsed_time": "1 day, 9:28:54", "remaining_time": "22:16:50"} +{"current_steps": 4130, "total_steps": 6870, "loss": 0.271, "lr": 2.0571256269500893e-05, "epoch": 1.8034934497816595, "percentage": 60.12, "elapsed_time": "1 day, 9:31:20", "remaining_time": "22:14:24"} +{"current_steps": 4135, "total_steps": 6870, "loss": 0.2683, "lr": 2.0508762442180745e-05, "epoch": 1.8056768558951966, "percentage": 60.19, "elapsed_time": "1 day, 9:33:46", "remaining_time": "22:11:58"} +{"current_steps": 4140, "total_steps": 6870, "loss": 0.2782, "lr": 2.0446297602113167e-05, "epoch": 1.8078602620087336, "percentage": 60.26, "elapsed_time": "1 day, 9:36:13", "remaining_time": "22:09:32"} +{"current_steps": 4145, "total_steps": 6870, "loss": 0.2636, "lr": 2.0383862152457432e-05, "epoch": 1.8100436681222707, "percentage": 60.33, "elapsed_time": "1 day, 9:38:39", "remaining_time": "22:07:06"} +{"current_steps": 4150, "total_steps": 6870, "loss": 0.276, "lr": 2.0321456496183134e-05, "epoch": 1.8122270742358078, "percentage": 60.41, "elapsed_time": "1 day, 9:41:05", "remaining_time": "22:04:39"} +{"current_steps": 4155, "total_steps": 6870, "loss": 0.2706, "lr": 2.0259081036067563e-05, "epoch": 1.8144104803493448, "percentage": 60.48, "elapsed_time": "1 day, 9:43:31", "remaining_time": "22:02:13"} +{"current_steps": 4160, "total_steps": 6870, "loss": 0.2529, "lr": 2.0196736174693126e-05, "epoch": 1.8165938864628821, "percentage": 60.55, "elapsed_time": "1 day, 9:45:57", "remaining_time": "21:59:47"} +{"current_steps": 4165, "total_steps": 6870, "loss": 0.2649, "lr": 2.013442231444474e-05, "epoch": 1.8187772925764192, "percentage": 60.63, "elapsed_time": "1 day, 9:48:24", "remaining_time": "21:57:21"} +{"current_steps": 4170, "total_steps": 6870, "loss": 0.265, "lr": 2.007213985750724e-05, "epoch": 1.8209606986899565, "percentage": 60.7, "elapsed_time": "1 day, 9:50:50", "remaining_time": "21:54:55"} +{"current_steps": 4175, "total_steps": 6870, "loss": 0.2653, "lr": 2.000988920586275e-05, "epoch": 1.8231441048034935, "percentage": 60.77, "elapsed_time": "1 day, 9:53:16", "remaining_time": "21:52:29"} +{"current_steps": 4180, "total_steps": 6870, "loss": 0.2572, "lr": 1.994767076128816e-05, "epoch": 1.8253275109170306, "percentage": 60.84, "elapsed_time": "1 day, 9:55:42", "remaining_time": "21:50:03"} +{"current_steps": 4185, "total_steps": 6870, "loss": 0.2754, "lr": 1.988548492535246e-05, "epoch": 1.8275109170305677, "percentage": 60.92, "elapsed_time": "1 day, 9:58:09", "remaining_time": "21:47:37"} +{"current_steps": 4190, "total_steps": 6870, "loss": 0.2717, "lr": 1.9823332099414185e-05, "epoch": 1.8296943231441047, "percentage": 60.99, "elapsed_time": "1 day, 10:00:35", "remaining_time": "21:45:11"} +{"current_steps": 4195, "total_steps": 6870, "loss": 0.25, "lr": 1.9761212684618824e-05, "epoch": 1.8318777292576418, "percentage": 61.06, "elapsed_time": "1 day, 10:03:01", "remaining_time": "21:42:45"} +{"current_steps": 4200, "total_steps": 6870, "loss": 0.2513, "lr": 1.9699127081896214e-05, "epoch": 1.8340611353711789, "percentage": 61.14, "elapsed_time": "1 day, 10:05:27", "remaining_time": "21:40:19"} +{"current_steps": 4205, "total_steps": 6870, "loss": 0.2524, "lr": 1.963707569195797e-05, "epoch": 1.8362445414847162, "percentage": 61.21, "elapsed_time": "1 day, 10:07:53", "remaining_time": "21:37:53"} +{"current_steps": 4210, "total_steps": 6870, "loss": 0.2652, "lr": 1.9575058915294885e-05, "epoch": 1.8384279475982532, "percentage": 61.28, "elapsed_time": "1 day, 10:10:19", "remaining_time": "21:35:27"} +{"current_steps": 4215, "total_steps": 6870, "loss": 0.2776, "lr": 1.9513077152174363e-05, "epoch": 1.8406113537117905, "percentage": 61.35, "elapsed_time": "1 day, 10:12:46", "remaining_time": "21:33:01"} +{"current_steps": 4220, "total_steps": 6870, "loss": 0.2588, "lr": 1.9451130802637822e-05, "epoch": 1.8427947598253276, "percentage": 61.43, "elapsed_time": "1 day, 10:15:12", "remaining_time": "21:30:35"} +{"current_steps": 4225, "total_steps": 6870, "loss": 0.2701, "lr": 1.9389220266498102e-05, "epoch": 1.8449781659388647, "percentage": 61.5, "elapsed_time": "1 day, 10:17:38", "remaining_time": "21:28:09"} +{"current_steps": 4230, "total_steps": 6870, "loss": 0.2645, "lr": 1.9327345943336904e-05, "epoch": 1.8471615720524017, "percentage": 61.57, "elapsed_time": "1 day, 10:20:04", "remaining_time": "21:25:43"} +{"current_steps": 4235, "total_steps": 6870, "loss": 0.2749, "lr": 1.9265508232502208e-05, "epoch": 1.8493449781659388, "percentage": 61.64, "elapsed_time": "1 day, 10:22:30", "remaining_time": "21:23:17"} +{"current_steps": 4240, "total_steps": 6870, "loss": 0.2598, "lr": 1.9203707533105687e-05, "epoch": 1.8515283842794759, "percentage": 61.72, "elapsed_time": "1 day, 10:24:56", "remaining_time": "21:20:51"} +{"current_steps": 4245, "total_steps": 6870, "loss": 0.271, "lr": 1.914194424402015e-05, "epoch": 1.8537117903930131, "percentage": 61.79, "elapsed_time": "1 day, 10:27:23", "remaining_time": "21:18:25"} +{"current_steps": 4250, "total_steps": 6870, "loss": 0.2512, "lr": 1.9080218763876917e-05, "epoch": 1.8558951965065502, "percentage": 61.86, "elapsed_time": "1 day, 10:29:49", "remaining_time": "21:15:58"} +{"current_steps": 4255, "total_steps": 6870, "loss": 0.263, "lr": 1.901853149106332e-05, "epoch": 1.8580786026200873, "percentage": 61.94, "elapsed_time": "1 day, 10:32:15", "remaining_time": "21:13:32"} +{"current_steps": 4260, "total_steps": 6870, "loss": 0.2657, "lr": 1.8956882823720075e-05, "epoch": 1.8602620087336246, "percentage": 62.01, "elapsed_time": "1 day, 10:34:41", "remaining_time": "21:11:06"} +{"current_steps": 4265, "total_steps": 6870, "loss": 0.26, "lr": 1.889527315973874e-05, "epoch": 1.8624454148471616, "percentage": 62.08, "elapsed_time": "1 day, 10:37:07", "remaining_time": "21:08:40"} +{"current_steps": 4270, "total_steps": 6870, "loss": 0.2636, "lr": 1.8833702896759132e-05, "epoch": 1.8646288209606987, "percentage": 62.15, "elapsed_time": "1 day, 10:39:33", "remaining_time": "21:06:14"} +{"current_steps": 4275, "total_steps": 6870, "loss": 0.2744, "lr": 1.8772172432166767e-05, "epoch": 1.8668122270742358, "percentage": 62.23, "elapsed_time": "1 day, 10:42:00", "remaining_time": "21:03:48"} +{"current_steps": 4280, "total_steps": 6870, "loss": 0.2748, "lr": 1.87106821630903e-05, "epoch": 1.8689956331877728, "percentage": 62.3, "elapsed_time": "1 day, 10:44:26", "remaining_time": "21:01:22"} +{"current_steps": 4285, "total_steps": 6870, "loss": 0.2638, "lr": 1.8649232486398955e-05, "epoch": 1.87117903930131, "percentage": 62.37, "elapsed_time": "1 day, 10:46:52", "remaining_time": "20:58:56"} +{"current_steps": 4290, "total_steps": 6870, "loss": 0.2601, "lr": 1.8587823798699958e-05, "epoch": 1.8733624454148472, "percentage": 62.45, "elapsed_time": "1 day, 10:49:18", "remaining_time": "20:56:30"} +{"current_steps": 4295, "total_steps": 6870, "loss": 0.2525, "lr": 1.8526456496335997e-05, "epoch": 1.8755458515283843, "percentage": 62.52, "elapsed_time": "1 day, 10:51:44", "remaining_time": "20:54:04"} +{"current_steps": 4300, "total_steps": 6870, "loss": 0.2547, "lr": 1.846513097538265e-05, "epoch": 1.8777292576419216, "percentage": 62.59, "elapsed_time": "1 day, 10:54:10", "remaining_time": "20:51:38"} +{"current_steps": 4305, "total_steps": 6870, "loss": 0.2747, "lr": 1.8403847631645815e-05, "epoch": 1.8799126637554586, "percentage": 62.66, "elapsed_time": "1 day, 10:56:36", "remaining_time": "20:49:12"} +{"current_steps": 4310, "total_steps": 6870, "loss": 0.2755, "lr": 1.834260686065919e-05, "epoch": 1.8820960698689957, "percentage": 62.74, "elapsed_time": "1 day, 10:59:02", "remaining_time": "20:46:46"} +{"current_steps": 4315, "total_steps": 6870, "loss": 0.2556, "lr": 1.8281409057681685e-05, "epoch": 1.8842794759825328, "percentage": 62.81, "elapsed_time": "1 day, 11:01:29", "remaining_time": "20:44:19"} +{"current_steps": 4320, "total_steps": 6870, "loss": 0.2725, "lr": 1.8220254617694904e-05, "epoch": 1.8864628820960698, "percentage": 62.88, "elapsed_time": "1 day, 11:03:55", "remaining_time": "20:41:53"} +{"current_steps": 4325, "total_steps": 6870, "loss": 0.2672, "lr": 1.815914393540057e-05, "epoch": 1.888646288209607, "percentage": 62.95, "elapsed_time": "1 day, 11:06:21", "remaining_time": "20:39:27"} +{"current_steps": 4330, "total_steps": 6870, "loss": 0.2681, "lr": 1.8098077405217973e-05, "epoch": 1.890829694323144, "percentage": 63.03, "elapsed_time": "1 day, 11:08:47", "remaining_time": "20:37:01"} +{"current_steps": 4335, "total_steps": 6870, "loss": 0.2754, "lr": 1.8037055421281453e-05, "epoch": 1.8930131004366813, "percentage": 63.1, "elapsed_time": "1 day, 11:11:13", "remaining_time": "20:34:35"} +{"current_steps": 4340, "total_steps": 6870, "loss": 0.254, "lr": 1.7976078377437843e-05, "epoch": 1.8951965065502183, "percentage": 63.17, "elapsed_time": "1 day, 11:13:39", "remaining_time": "20:32:09"} +{"current_steps": 4345, "total_steps": 6870, "loss": 0.2709, "lr": 1.7915146667243912e-05, "epoch": 1.8973799126637556, "percentage": 63.25, "elapsed_time": "1 day, 11:16:05", "remaining_time": "20:29:43"} +{"current_steps": 4350, "total_steps": 6870, "loss": 0.2724, "lr": 1.785426068396385e-05, "epoch": 1.8995633187772927, "percentage": 63.32, "elapsed_time": "1 day, 11:18:31", "remaining_time": "20:27:17"} +{"current_steps": 4355, "total_steps": 6870, "loss": 0.2685, "lr": 1.7793420820566698e-05, "epoch": 1.9017467248908297, "percentage": 63.39, "elapsed_time": "1 day, 11:20:57", "remaining_time": "20:24:51"} +{"current_steps": 4360, "total_steps": 6870, "loss": 0.2665, "lr": 1.7732627469723867e-05, "epoch": 1.9039301310043668, "percentage": 63.46, "elapsed_time": "1 day, 11:23:23", "remaining_time": "20:22:24"} +{"current_steps": 4365, "total_steps": 6870, "loss": 0.2629, "lr": 1.7671881023806526e-05, "epoch": 1.9061135371179039, "percentage": 63.54, "elapsed_time": "1 day, 11:25:49", "remaining_time": "20:19:58"} +{"current_steps": 4370, "total_steps": 6870, "loss": 0.2637, "lr": 1.7611181874883133e-05, "epoch": 1.908296943231441, "percentage": 63.61, "elapsed_time": "1 day, 11:28:16", "remaining_time": "20:17:32"} +{"current_steps": 4375, "total_steps": 6870, "loss": 0.2626, "lr": 1.755053041471689e-05, "epoch": 1.910480349344978, "percentage": 63.68, "elapsed_time": "1 day, 11:30:42", "remaining_time": "20:15:06"} +{"current_steps": 4380, "total_steps": 6870, "loss": 0.2621, "lr": 1.748992703476318e-05, "epoch": 1.9126637554585153, "percentage": 63.76, "elapsed_time": "1 day, 11:33:08", "remaining_time": "20:12:40"} +{"current_steps": 4385, "total_steps": 6870, "loss": 0.2492, "lr": 1.7429372126167105e-05, "epoch": 1.9148471615720524, "percentage": 63.83, "elapsed_time": "1 day, 11:35:34", "remaining_time": "20:10:14"} +{"current_steps": 4390, "total_steps": 6870, "loss": 0.2615, "lr": 1.7368866079760887e-05, "epoch": 1.9170305676855897, "percentage": 63.9, "elapsed_time": "1 day, 11:38:00", "remaining_time": "20:07:48"} +{"current_steps": 4395, "total_steps": 6870, "loss": 0.2686, "lr": 1.7308409286061407e-05, "epoch": 1.9192139737991267, "percentage": 63.97, "elapsed_time": "1 day, 11:40:26", "remaining_time": "20:05:22"} +{"current_steps": 4400, "total_steps": 6870, "loss": 0.2666, "lr": 1.7248002135267645e-05, "epoch": 1.9213973799126638, "percentage": 64.05, "elapsed_time": "1 day, 11:42:53", "remaining_time": "20:02:56"} +{"current_steps": 4405, "total_steps": 6870, "loss": 0.2735, "lr": 1.7187645017258196e-05, "epoch": 1.9235807860262009, "percentage": 64.12, "elapsed_time": "1 day, 11:45:19", "remaining_time": "20:00:30"} +{"current_steps": 4410, "total_steps": 6870, "loss": 0.2613, "lr": 1.712733832158872e-05, "epoch": 1.925764192139738, "percentage": 64.19, "elapsed_time": "1 day, 11:47:45", "remaining_time": "19:58:04"} +{"current_steps": 4415, "total_steps": 6870, "loss": 0.2653, "lr": 1.7067082437489436e-05, "epoch": 1.927947598253275, "percentage": 64.26, "elapsed_time": "1 day, 11:50:11", "remaining_time": "19:55:38"} +{"current_steps": 4420, "total_steps": 6870, "loss": 0.262, "lr": 1.7006877753862632e-05, "epoch": 1.930131004366812, "percentage": 64.34, "elapsed_time": "1 day, 11:52:37", "remaining_time": "19:53:12"} +{"current_steps": 4425, "total_steps": 6870, "loss": 0.2542, "lr": 1.6946724659280133e-05, "epoch": 1.9323144104803494, "percentage": 64.41, "elapsed_time": "1 day, 11:55:04", "remaining_time": "19:50:45"} +{"current_steps": 4430, "total_steps": 6870, "loss": 0.2757, "lr": 1.6886623541980796e-05, "epoch": 1.9344978165938864, "percentage": 64.48, "elapsed_time": "1 day, 11:57:30", "remaining_time": "19:48:19"} +{"current_steps": 4435, "total_steps": 6870, "loss": 0.2645, "lr": 1.682657478986801e-05, "epoch": 1.9366812227074237, "percentage": 64.56, "elapsed_time": "1 day, 11:59:56", "remaining_time": "19:45:53"} +{"current_steps": 4440, "total_steps": 6870, "loss": 0.2595, "lr": 1.6766578790507192e-05, "epoch": 1.9388646288209608, "percentage": 64.63, "elapsed_time": "1 day, 12:02:22", "remaining_time": "19:43:27"} +{"current_steps": 4445, "total_steps": 6870, "loss": 0.2596, "lr": 1.6706635931123277e-05, "epoch": 1.9410480349344978, "percentage": 64.7, "elapsed_time": "1 day, 12:04:48", "remaining_time": "19:41:01"} +{"current_steps": 4450, "total_steps": 6870, "loss": 0.2493, "lr": 1.664674659859822e-05, "epoch": 1.943231441048035, "percentage": 64.77, "elapsed_time": "1 day, 12:07:14", "remaining_time": "19:38:35"} +{"current_steps": 4455, "total_steps": 6870, "loss": 0.2761, "lr": 1.658691117946853e-05, "epoch": 1.945414847161572, "percentage": 64.85, "elapsed_time": "1 day, 12:09:40", "remaining_time": "19:36:09"} +{"current_steps": 4460, "total_steps": 6870, "loss": 0.2632, "lr": 1.6527130059922717e-05, "epoch": 1.947598253275109, "percentage": 64.92, "elapsed_time": "1 day, 12:12:07", "remaining_time": "19:33:43"} +{"current_steps": 4465, "total_steps": 6870, "loss": 0.2615, "lr": 1.646740362579886e-05, "epoch": 1.9497816593886463, "percentage": 64.99, "elapsed_time": "1 day, 12:14:33", "remaining_time": "19:31:17"} +{"current_steps": 4470, "total_steps": 6870, "loss": 0.2654, "lr": 1.6407732262582067e-05, "epoch": 1.9519650655021834, "percentage": 65.07, "elapsed_time": "1 day, 12:16:59", "remaining_time": "19:28:51"} +{"current_steps": 4475, "total_steps": 6870, "loss": 0.2593, "lr": 1.634811635540201e-05, "epoch": 1.9541484716157205, "percentage": 65.14, "elapsed_time": "1 day, 12:19:25", "remaining_time": "19:26:25"} +{"current_steps": 4480, "total_steps": 6870, "loss": 0.2676, "lr": 1.6288556289030452e-05, "epoch": 1.9563318777292578, "percentage": 65.21, "elapsed_time": "1 day, 12:21:51", "remaining_time": "19:23:59"} +{"current_steps": 4485, "total_steps": 6870, "loss": 0.2541, "lr": 1.6229052447878743e-05, "epoch": 1.9585152838427948, "percentage": 65.28, "elapsed_time": "1 day, 12:24:17", "remaining_time": "19:21:32"} +{"current_steps": 4490, "total_steps": 6870, "loss": 0.259, "lr": 1.6169605215995345e-05, "epoch": 1.960698689956332, "percentage": 65.36, "elapsed_time": "1 day, 12:26:43", "remaining_time": "19:19:06"} +{"current_steps": 4495, "total_steps": 6870, "loss": 0.2559, "lr": 1.6110214977063345e-05, "epoch": 1.962882096069869, "percentage": 65.43, "elapsed_time": "1 day, 12:29:09", "remaining_time": "19:16:40"} +{"current_steps": 4500, "total_steps": 6870, "loss": 0.2508, "lr": 1.6050882114398e-05, "epoch": 1.965065502183406, "percentage": 65.5, "elapsed_time": "1 day, 12:31:36", "remaining_time": "19:14:14"} +{"current_steps": 4505, "total_steps": 6870, "loss": 0.2607, "lr": 1.5991607010944242e-05, "epoch": 1.967248908296943, "percentage": 65.57, "elapsed_time": "1 day, 12:34:02", "remaining_time": "19:11:48"} +{"current_steps": 4510, "total_steps": 6870, "loss": 0.2644, "lr": 1.5932390049274227e-05, "epoch": 1.9694323144104804, "percentage": 65.65, "elapsed_time": "1 day, 12:36:28", "remaining_time": "19:09:22"} +{"current_steps": 4515, "total_steps": 6870, "loss": 0.272, "lr": 1.587323161158484e-05, "epoch": 1.9716157205240175, "percentage": 65.72, "elapsed_time": "1 day, 12:38:54", "remaining_time": "19:06:56"} +{"current_steps": 4520, "total_steps": 6870, "loss": 0.2649, "lr": 1.5814132079695232e-05, "epoch": 1.9737991266375547, "percentage": 65.79, "elapsed_time": "1 day, 12:41:20", "remaining_time": "19:04:30"} +{"current_steps": 4525, "total_steps": 6870, "loss": 0.2591, "lr": 1.57550918350444e-05, "epoch": 1.9759825327510918, "percentage": 65.87, "elapsed_time": "1 day, 12:43:47", "remaining_time": "19:02:04"} +{"current_steps": 4530, "total_steps": 6870, "loss": 0.256, "lr": 1.5696111258688657e-05, "epoch": 1.9781659388646289, "percentage": 65.94, "elapsed_time": "1 day, 12:46:13", "remaining_time": "18:59:38"} +{"current_steps": 4535, "total_steps": 6870, "loss": 0.2655, "lr": 1.563719073129923e-05, "epoch": 1.980349344978166, "percentage": 66.01, "elapsed_time": "1 day, 12:48:39", "remaining_time": "18:57:12"} +{"current_steps": 4540, "total_steps": 6870, "loss": 0.2593, "lr": 1.557833063315976e-05, "epoch": 1.982532751091703, "percentage": 66.08, "elapsed_time": "1 day, 12:51:05", "remaining_time": "18:54:46"} +{"current_steps": 4545, "total_steps": 6870, "loss": 0.2593, "lr": 1.5519531344163873e-05, "epoch": 1.98471615720524, "percentage": 66.16, "elapsed_time": "1 day, 12:53:31", "remaining_time": "18:52:20"} +{"current_steps": 4550, "total_steps": 6870, "loss": 0.2645, "lr": 1.5460793243812732e-05, "epoch": 1.9868995633187772, "percentage": 66.23, "elapsed_time": "1 day, 12:55:58", "remaining_time": "18:49:54"} +{"current_steps": 4555, "total_steps": 6870, "loss": 0.2712, "lr": 1.5402116711212566e-05, "epoch": 1.9890829694323144, "percentage": 66.3, "elapsed_time": "1 day, 12:58:24", "remaining_time": "18:47:27"} +{"current_steps": 4560, "total_steps": 6870, "loss": 0.2542, "lr": 1.5343502125072236e-05, "epoch": 1.9912663755458515, "percentage": 66.38, "elapsed_time": "1 day, 13:00:50", "remaining_time": "18:45:01"} +{"current_steps": 4565, "total_steps": 6870, "loss": 0.2663, "lr": 1.5284949863700805e-05, "epoch": 1.9934497816593888, "percentage": 66.45, "elapsed_time": "1 day, 13:03:16", "remaining_time": "18:42:35"} +{"current_steps": 4570, "total_steps": 6870, "loss": 0.2622, "lr": 1.5226460305005044e-05, "epoch": 1.9956331877729259, "percentage": 66.52, "elapsed_time": "1 day, 13:05:42", "remaining_time": "18:40:09"} +{"current_steps": 4575, "total_steps": 6870, "loss": 0.2692, "lr": 1.5168033826487072e-05, "epoch": 1.997816593886463, "percentage": 66.59, "elapsed_time": "1 day, 13:08:09", "remaining_time": "18:37:43"} +{"current_steps": 4580, "total_steps": 6870, "loss": 0.2572, "lr": 1.5109670805241846e-05, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "1 day, 13:10:35", "remaining_time": "18:35:17"} +{"current_steps": 4585, "total_steps": 6870, "loss": 0.2211, "lr": 1.505137161795478e-05, "epoch": 2.002183406113537, "percentage": 66.74, "elapsed_time": "1 day, 13:13:01", "remaining_time": "18:32:51"} +{"current_steps": 4590, "total_steps": 6870, "loss": 0.2147, "lr": 1.4993136640899286e-05, "epoch": 2.004366812227074, "percentage": 66.81, "elapsed_time": "1 day, 13:15:27", "remaining_time": "18:30:25"} +{"current_steps": 4595, "total_steps": 6870, "loss": 0.2052, "lr": 1.4934966249934334e-05, "epoch": 2.006550218340611, "percentage": 66.89, "elapsed_time": "1 day, 13:17:53", "remaining_time": "18:27:59"} +{"current_steps": 4600, "total_steps": 6870, "loss": 0.2162, "lr": 1.487686082050207e-05, "epoch": 2.0087336244541483, "percentage": 66.96, "elapsed_time": "1 day, 13:20:20", "remaining_time": "18:25:33"} +{"current_steps": 4605, "total_steps": 6870, "loss": 0.2071, "lr": 1.4818820727625348e-05, "epoch": 2.010917030567686, "percentage": 67.03, "elapsed_time": "1 day, 13:22:46", "remaining_time": "18:23:07"} +{"current_steps": 4610, "total_steps": 6870, "loss": 0.2139, "lr": 1.4760846345905344e-05, "epoch": 2.013100436681223, "percentage": 67.1, "elapsed_time": "1 day, 13:25:12", "remaining_time": "18:20:41"} +{"current_steps": 4615, "total_steps": 6870, "loss": 0.2168, "lr": 1.4702938049519105e-05, "epoch": 2.01528384279476, "percentage": 67.18, "elapsed_time": "1 day, 13:27:38", "remaining_time": "18:18:15"} +{"current_steps": 4620, "total_steps": 6870, "loss": 0.2073, "lr": 1.4645096212217167e-05, "epoch": 2.017467248908297, "percentage": 67.25, "elapsed_time": "1 day, 13:30:04", "remaining_time": "18:15:49"} +{"current_steps": 4625, "total_steps": 6870, "loss": 0.2233, "lr": 1.4587321207321115e-05, "epoch": 2.019650655021834, "percentage": 67.32, "elapsed_time": "1 day, 13:32:30", "remaining_time": "18:13:23"} +{"current_steps": 4630, "total_steps": 6870, "loss": 0.2016, "lr": 1.4529613407721192e-05, "epoch": 2.021834061135371, "percentage": 67.39, "elapsed_time": "1 day, 13:34:57", "remaining_time": "18:10:56"} +{"current_steps": 4635, "total_steps": 6870, "loss": 0.2092, "lr": 1.4471973185873888e-05, "epoch": 2.024017467248908, "percentage": 67.47, "elapsed_time": "1 day, 13:37:23", "remaining_time": "18:08:30"} +{"current_steps": 4640, "total_steps": 6870, "loss": 0.2026, "lr": 1.4414400913799535e-05, "epoch": 2.0262008733624453, "percentage": 67.54, "elapsed_time": "1 day, 13:39:49", "remaining_time": "18:06:04"} +{"current_steps": 4645, "total_steps": 6870, "loss": 0.2034, "lr": 1.4356896963079885e-05, "epoch": 2.0283842794759823, "percentage": 67.61, "elapsed_time": "1 day, 13:42:15", "remaining_time": "18:03:38"} +{"current_steps": 4650, "total_steps": 6870, "loss": 0.2038, "lr": 1.4299461704855759e-05, "epoch": 2.03056768558952, "percentage": 67.69, "elapsed_time": "1 day, 13:44:41", "remaining_time": "18:01:12"} +{"current_steps": 4655, "total_steps": 6870, "loss": 0.2151, "lr": 1.424209550982461e-05, "epoch": 2.032751091703057, "percentage": 67.76, "elapsed_time": "1 day, 13:47:08", "remaining_time": "17:58:46"} +{"current_steps": 4660, "total_steps": 6870, "loss": 0.2176, "lr": 1.4184798748238148e-05, "epoch": 2.034934497816594, "percentage": 67.83, "elapsed_time": "1 day, 13:49:34", "remaining_time": "17:56:20"} +{"current_steps": 4665, "total_steps": 6870, "loss": 0.2077, "lr": 1.4127571789899946e-05, "epoch": 2.037117903930131, "percentage": 67.9, "elapsed_time": "1 day, 13:52:00", "remaining_time": "17:53:54"} +{"current_steps": 4670, "total_steps": 6870, "loss": 0.2163, "lr": 1.4070415004163056e-05, "epoch": 2.039301310043668, "percentage": 67.98, "elapsed_time": "1 day, 13:54:26", "remaining_time": "17:51:28"} +{"current_steps": 4675, "total_steps": 6870, "loss": 0.2139, "lr": 1.4013328759927625e-05, "epoch": 2.041484716157205, "percentage": 68.05, "elapsed_time": "1 day, 13:56:52", "remaining_time": "17:49:02"} +{"current_steps": 4680, "total_steps": 6870, "loss": 0.2019, "lr": 1.39563134256385e-05, "epoch": 2.0436681222707422, "percentage": 68.12, "elapsed_time": "1 day, 13:59:18", "remaining_time": "17:46:36"} +{"current_steps": 4685, "total_steps": 6870, "loss": 0.2183, "lr": 1.3899369369282888e-05, "epoch": 2.0458515283842793, "percentage": 68.2, "elapsed_time": "1 day, 14:01:44", "remaining_time": "17:44:09"} +{"current_steps": 4690, "total_steps": 6870, "loss": 0.2188, "lr": 1.3842496958387938e-05, "epoch": 2.0480349344978164, "percentage": 68.27, "elapsed_time": "1 day, 14:04:10", "remaining_time": "17:41:43"} +{"current_steps": 4695, "total_steps": 6870, "loss": 0.2038, "lr": 1.3785696560018374e-05, "epoch": 2.050218340611354, "percentage": 68.34, "elapsed_time": "1 day, 14:06:36", "remaining_time": "17:39:17"} +{"current_steps": 4700, "total_steps": 6870, "loss": 0.206, "lr": 1.372896854077416e-05, "epoch": 2.052401746724891, "percentage": 68.41, "elapsed_time": "1 day, 14:09:03", "remaining_time": "17:36:51"} +{"current_steps": 4705, "total_steps": 6870, "loss": 0.2133, "lr": 1.3672313266788106e-05, "epoch": 2.054585152838428, "percentage": 68.49, "elapsed_time": "1 day, 14:11:29", "remaining_time": "17:34:25"} +{"current_steps": 4710, "total_steps": 6870, "loss": 0.2126, "lr": 1.3615731103723507e-05, "epoch": 2.056768558951965, "percentage": 68.56, "elapsed_time": "1 day, 14:13:55", "remaining_time": "17:31:59"} +{"current_steps": 4715, "total_steps": 6870, "loss": 0.2061, "lr": 1.3559222416771793e-05, "epoch": 2.058951965065502, "percentage": 68.63, "elapsed_time": "1 day, 14:16:21", "remaining_time": "17:29:33"} +{"current_steps": 4720, "total_steps": 6870, "loss": 0.2109, "lr": 1.350278757065016e-05, "epoch": 2.061135371179039, "percentage": 68.7, "elapsed_time": "1 day, 14:18:47", "remaining_time": "17:27:07"} +{"current_steps": 4725, "total_steps": 6870, "loss": 0.2137, "lr": 1.3446426929599229e-05, "epoch": 2.0633187772925763, "percentage": 68.78, "elapsed_time": "1 day, 14:21:13", "remaining_time": "17:24:41"} +{"current_steps": 4730, "total_steps": 6870, "loss": 0.1967, "lr": 1.3390140857380678e-05, "epoch": 2.0655021834061134, "percentage": 68.85, "elapsed_time": "1 day, 14:23:40", "remaining_time": "17:22:15"} +{"current_steps": 4735, "total_steps": 6870, "loss": 0.2255, "lr": 1.3333929717274919e-05, "epoch": 2.067685589519651, "percentage": 68.92, "elapsed_time": "1 day, 14:26:06", "remaining_time": "17:19:48"} +{"current_steps": 4740, "total_steps": 6870, "loss": 0.2195, "lr": 1.3277793872078732e-05, "epoch": 2.069868995633188, "percentage": 69.0, "elapsed_time": "1 day, 14:28:32", "remaining_time": "17:17:22"} +{"current_steps": 4745, "total_steps": 6870, "loss": 0.2128, "lr": 1.3221733684102905e-05, "epoch": 2.072052401746725, "percentage": 69.07, "elapsed_time": "1 day, 14:30:58", "remaining_time": "17:14:56"} +{"current_steps": 4750, "total_steps": 6870, "loss": 0.2115, "lr": 1.3165749515169967e-05, "epoch": 2.074235807860262, "percentage": 69.14, "elapsed_time": "1 day, 14:33:24", "remaining_time": "17:12:30"} +{"current_steps": 4755, "total_steps": 6870, "loss": 0.2181, "lr": 1.310984172661177e-05, "epoch": 2.076419213973799, "percentage": 69.21, "elapsed_time": "1 day, 14:35:50", "remaining_time": "17:10:04"} +{"current_steps": 4760, "total_steps": 6870, "loss": 0.2143, "lr": 1.305401067926722e-05, "epoch": 2.078602620087336, "percentage": 69.29, "elapsed_time": "1 day, 14:38:17", "remaining_time": "17:07:38"} +{"current_steps": 4765, "total_steps": 6870, "loss": 0.2002, "lr": 1.2998256733479897e-05, "epoch": 2.0807860262008733, "percentage": 69.36, "elapsed_time": "1 day, 14:40:43", "remaining_time": "17:05:12"} +{"current_steps": 4770, "total_steps": 6870, "loss": 0.2041, "lr": 1.294258024909577e-05, "epoch": 2.0829694323144103, "percentage": 69.43, "elapsed_time": "1 day, 14:43:09", "remaining_time": "17:02:46"} +{"current_steps": 4775, "total_steps": 6870, "loss": 0.2206, "lr": 1.288698158546085e-05, "epoch": 2.0851528384279474, "percentage": 69.51, "elapsed_time": "1 day, 14:45:35", "remaining_time": "17:00:20"} +{"current_steps": 4780, "total_steps": 6870, "loss": 0.2166, "lr": 1.2831461101418884e-05, "epoch": 2.087336244541485, "percentage": 69.58, "elapsed_time": "1 day, 14:48:01", "remaining_time": "16:57:54"} +{"current_steps": 4785, "total_steps": 6870, "loss": 0.2228, "lr": 1.2776019155309035e-05, "epoch": 2.089519650655022, "percentage": 69.65, "elapsed_time": "1 day, 14:50:28", "remaining_time": "16:55:28"} +{"current_steps": 4790, "total_steps": 6870, "loss": 0.2146, "lr": 1.2720656104963569e-05, "epoch": 2.091703056768559, "percentage": 69.72, "elapsed_time": "1 day, 14:52:54", "remaining_time": "16:53:02"} +{"current_steps": 4795, "total_steps": 6870, "loss": 0.2173, "lr": 1.2665372307705525e-05, "epoch": 2.093886462882096, "percentage": 69.8, "elapsed_time": "1 day, 14:55:20", "remaining_time": "16:50:36"} +{"current_steps": 4800, "total_steps": 6870, "loss": 0.2203, "lr": 1.2610168120346455e-05, "epoch": 2.096069868995633, "percentage": 69.87, "elapsed_time": "1 day, 14:57:46", "remaining_time": "16:48:09"} +{"current_steps": 4805, "total_steps": 6870, "loss": 0.2169, "lr": 1.2555043899184077e-05, "epoch": 2.0982532751091703, "percentage": 69.94, "elapsed_time": "1 day, 15:00:12", "remaining_time": "16:45:43"} +{"current_steps": 4810, "total_steps": 6870, "loss": 0.2131, "lr": 1.2500000000000006e-05, "epoch": 2.1004366812227073, "percentage": 70.01, "elapsed_time": "1 day, 15:02:38", "remaining_time": "16:43:17"} +{"current_steps": 4815, "total_steps": 6870, "loss": 0.2155, "lr": 1.2445036778057434e-05, "epoch": 2.1026200873362444, "percentage": 70.09, "elapsed_time": "1 day, 15:05:05", "remaining_time": "16:40:51"} +{"current_steps": 4820, "total_steps": 6870, "loss": 0.2151, "lr": 1.2390154588098854e-05, "epoch": 2.1048034934497815, "percentage": 70.16, "elapsed_time": "1 day, 15:07:31", "remaining_time": "16:38:25"} +{"current_steps": 4825, "total_steps": 6870, "loss": 0.2075, "lr": 1.233535378434376e-05, "epoch": 2.106986899563319, "percentage": 70.23, "elapsed_time": "1 day, 15:09:57", "remaining_time": "16:35:59"} +{"current_steps": 4830, "total_steps": 6870, "loss": 0.2136, "lr": 1.2280634720486364e-05, "epoch": 2.109170305676856, "percentage": 70.31, "elapsed_time": "1 day, 15:12:23", "remaining_time": "16:33:33"} +{"current_steps": 4835, "total_steps": 6870, "loss": 0.21, "lr": 1.2225997749693318e-05, "epoch": 2.111353711790393, "percentage": 70.38, "elapsed_time": "1 day, 15:14:49", "remaining_time": "16:31:07"} +{"current_steps": 4840, "total_steps": 6870, "loss": 0.2141, "lr": 1.2171443224601443e-05, "epoch": 2.11353711790393, "percentage": 70.45, "elapsed_time": "1 day, 15:17:15", "remaining_time": "16:28:41"} +{"current_steps": 4845, "total_steps": 6870, "loss": 0.208, "lr": 1.2116971497315402e-05, "epoch": 2.1157205240174672, "percentage": 70.52, "elapsed_time": "1 day, 15:19:42", "remaining_time": "16:26:15"} +{"current_steps": 4850, "total_steps": 6870, "loss": 0.2093, "lr": 1.2062582919405502e-05, "epoch": 2.1179039301310043, "percentage": 70.6, "elapsed_time": "1 day, 15:22:08", "remaining_time": "16:23:49"} +{"current_steps": 4855, "total_steps": 6870, "loss": 0.2146, "lr": 1.200827784190537e-05, "epoch": 2.1200873362445414, "percentage": 70.67, "elapsed_time": "1 day, 15:24:34", "remaining_time": "16:21:23"} +{"current_steps": 4860, "total_steps": 6870, "loss": 0.2084, "lr": 1.1954056615309728e-05, "epoch": 2.1222707423580784, "percentage": 70.74, "elapsed_time": "1 day, 15:27:00", "remaining_time": "16:18:56"} +{"current_steps": 4865, "total_steps": 6870, "loss": 0.2126, "lr": 1.1899919589572092e-05, "epoch": 2.1244541484716155, "percentage": 70.82, "elapsed_time": "1 day, 15:29:26", "remaining_time": "16:16:30"} +{"current_steps": 4870, "total_steps": 6870, "loss": 0.2063, "lr": 1.1845867114102532e-05, "epoch": 2.126637554585153, "percentage": 70.89, "elapsed_time": "1 day, 15:31:53", "remaining_time": "16:14:04"} +{"current_steps": 4875, "total_steps": 6870, "loss": 0.2224, "lr": 1.1791899537765416e-05, "epoch": 2.12882096069869, "percentage": 70.96, "elapsed_time": "1 day, 15:34:19", "remaining_time": "16:11:38"} +{"current_steps": 4880, "total_steps": 6870, "loss": 0.2218, "lr": 1.173801720887717e-05, "epoch": 2.131004366812227, "percentage": 71.03, "elapsed_time": "1 day, 15:36:45", "remaining_time": "16:09:12"} +{"current_steps": 4885, "total_steps": 6870, "loss": 0.199, "lr": 1.1684220475203986e-05, "epoch": 2.1331877729257642, "percentage": 71.11, "elapsed_time": "1 day, 15:39:11", "remaining_time": "16:06:46"} +{"current_steps": 4890, "total_steps": 6870, "loss": 0.2121, "lr": 1.1630509683959643e-05, "epoch": 2.1353711790393013, "percentage": 71.18, "elapsed_time": "1 day, 15:41:37", "remaining_time": "16:04:20"} +{"current_steps": 4895, "total_steps": 6870, "loss": 0.2171, "lr": 1.1576885181803215e-05, "epoch": 2.1375545851528384, "percentage": 71.25, "elapsed_time": "1 day, 15:44:03", "remaining_time": "16:01:54"} +{"current_steps": 4900, "total_steps": 6870, "loss": 0.2135, "lr": 1.1523347314836858e-05, "epoch": 2.1397379912663754, "percentage": 71.32, "elapsed_time": "1 day, 15:46:30", "remaining_time": "15:59:28"} +{"current_steps": 4905, "total_steps": 6870, "loss": 0.22, "lr": 1.1469896428603563e-05, "epoch": 2.1419213973799125, "percentage": 71.4, "elapsed_time": "1 day, 15:48:57", "remaining_time": "15:57:02"} +{"current_steps": 4910, "total_steps": 6870, "loss": 0.2022, "lr": 1.1416532868084937e-05, "epoch": 2.14410480349345, "percentage": 71.47, "elapsed_time": "1 day, 15:51:23", "remaining_time": "15:54:36"} +{"current_steps": 4915, "total_steps": 6870, "loss": 0.2157, "lr": 1.1363256977698968e-05, "epoch": 2.146288209606987, "percentage": 71.54, "elapsed_time": "1 day, 15:53:50", "remaining_time": "15:52:10"} +{"current_steps": 4920, "total_steps": 6870, "loss": 0.212, "lr": 1.1310069101297807e-05, "epoch": 2.148471615720524, "percentage": 71.62, "elapsed_time": "1 day, 15:56:16", "remaining_time": "15:49:44"} +{"current_steps": 4925, "total_steps": 6870, "loss": 0.21, "lr": 1.1256969582165544e-05, "epoch": 2.150655021834061, "percentage": 71.69, "elapsed_time": "1 day, 15:58:43", "remaining_time": "15:47:18"} +{"current_steps": 4930, "total_steps": 6870, "loss": 0.1969, "lr": 1.1203958763016007e-05, "epoch": 2.1528384279475983, "percentage": 71.76, "elapsed_time": "1 day, 16:01:09", "remaining_time": "15:44:52"} +{"current_steps": 4935, "total_steps": 6870, "loss": 0.209, "lr": 1.1151036985990509e-05, "epoch": 2.1550218340611353, "percentage": 71.83, "elapsed_time": "1 day, 16:03:35", "remaining_time": "15:42:26"} +{"current_steps": 4940, "total_steps": 6870, "loss": 0.2144, "lr": 1.1098204592655703e-05, "epoch": 2.1572052401746724, "percentage": 71.91, "elapsed_time": "1 day, 16:06:01", "remaining_time": "15:40:00"} +{"current_steps": 4945, "total_steps": 6870, "loss": 0.2095, "lr": 1.1045461924001325e-05, "epoch": 2.1593886462882095, "percentage": 71.98, "elapsed_time": "1 day, 16:08:27", "remaining_time": "15:37:34"} +{"current_steps": 4950, "total_steps": 6870, "loss": 0.2071, "lr": 1.0992809320438016e-05, "epoch": 2.1615720524017465, "percentage": 72.05, "elapsed_time": "1 day, 16:10:53", "remaining_time": "15:35:08"} +{"current_steps": 4955, "total_steps": 6870, "loss": 0.2095, "lr": 1.0940247121795123e-05, "epoch": 2.1637554585152836, "percentage": 72.13, "elapsed_time": "1 day, 16:13:19", "remaining_time": "15:32:41"} +{"current_steps": 4960, "total_steps": 6870, "loss": 0.2174, "lr": 1.08877756673185e-05, "epoch": 2.165938864628821, "percentage": 72.2, "elapsed_time": "1 day, 16:15:45", "remaining_time": "15:30:15"} +{"current_steps": 4965, "total_steps": 6870, "loss": 0.2179, "lr": 1.0835395295668327e-05, "epoch": 2.168122270742358, "percentage": 72.27, "elapsed_time": "1 day, 16:18:12", "remaining_time": "15:27:49"} +{"current_steps": 4970, "total_steps": 6870, "loss": 0.2051, "lr": 1.0783106344916913e-05, "epoch": 2.1703056768558953, "percentage": 72.34, "elapsed_time": "1 day, 16:20:38", "remaining_time": "15:25:23"} +{"current_steps": 4975, "total_steps": 6870, "loss": 0.2221, "lr": 1.073090915254652e-05, "epoch": 2.1724890829694323, "percentage": 72.42, "elapsed_time": "1 day, 16:23:04", "remaining_time": "15:22:57"} +{"current_steps": 4980, "total_steps": 6870, "loss": 0.2003, "lr": 1.06788040554472e-05, "epoch": 2.1746724890829694, "percentage": 72.49, "elapsed_time": "1 day, 16:25:30", "remaining_time": "15:20:31"} +{"current_steps": 4985, "total_steps": 6870, "loss": 0.2029, "lr": 1.0626791389914573e-05, "epoch": 2.1768558951965065, "percentage": 72.56, "elapsed_time": "1 day, 16:27:56", "remaining_time": "15:18:05"} +{"current_steps": 4990, "total_steps": 6870, "loss": 0.2141, "lr": 1.0574871491647723e-05, "epoch": 2.1790393013100435, "percentage": 72.63, "elapsed_time": "1 day, 16:30:22", "remaining_time": "15:15:39"} +{"current_steps": 4995, "total_steps": 6870, "loss": 0.2121, "lr": 1.0523044695746975e-05, "epoch": 2.1812227074235806, "percentage": 72.71, "elapsed_time": "1 day, 16:32:48", "remaining_time": "15:13:13"} +{"current_steps": 5000, "total_steps": 6870, "loss": 0.2123, "lr": 1.0471311336711772e-05, "epoch": 2.183406113537118, "percentage": 72.78, "elapsed_time": "1 day, 16:35:15", "remaining_time": "15:10:47"} +{"current_steps": 5005, "total_steps": 6870, "loss": 0.2166, "lr": 1.0419671748438487e-05, "epoch": 2.185589519650655, "percentage": 72.85, "elapsed_time": "1 day, 16:37:40", "remaining_time": "15:08:20"} +{"current_steps": 5010, "total_steps": 6870, "loss": 0.2109, "lr": 1.0368126264218282e-05, "epoch": 2.1877729257641922, "percentage": 72.93, "elapsed_time": "1 day, 16:40:06", "remaining_time": "15:05:54"} +{"current_steps": 5015, "total_steps": 6870, "loss": 0.2057, "lr": 1.0316675216734952e-05, "epoch": 2.1899563318777293, "percentage": 73.0, "elapsed_time": "1 day, 16:42:32", "remaining_time": "15:03:28"} +{"current_steps": 5020, "total_steps": 6870, "loss": 0.2225, "lr": 1.026531893806279e-05, "epoch": 2.1921397379912664, "percentage": 73.07, "elapsed_time": "1 day, 16:44:59", "remaining_time": "15:01:02"} +{"current_steps": 5025, "total_steps": 6870, "loss": 0.2161, "lr": 1.0214057759664424e-05, "epoch": 2.1943231441048034, "percentage": 73.14, "elapsed_time": "1 day, 16:47:25", "remaining_time": "14:58:36"} +{"current_steps": 5030, "total_steps": 6870, "loss": 0.223, "lr": 1.0162892012388703e-05, "epoch": 2.1965065502183405, "percentage": 73.22, "elapsed_time": "1 day, 16:49:51", "remaining_time": "14:56:10"} +{"current_steps": 5035, "total_steps": 6870, "loss": 0.2127, "lr": 1.0111822026468515e-05, "epoch": 2.1986899563318776, "percentage": 73.29, "elapsed_time": "1 day, 16:52:17", "remaining_time": "14:53:44"} +{"current_steps": 5040, "total_steps": 6870, "loss": 0.2132, "lr": 1.0060848131518716e-05, "epoch": 2.2008733624454146, "percentage": 73.36, "elapsed_time": "1 day, 16:54:43", "remaining_time": "14:51:18"} +{"current_steps": 5045, "total_steps": 6870, "loss": 0.2138, "lr": 1.0009970656533965e-05, "epoch": 2.203056768558952, "percentage": 73.44, "elapsed_time": "1 day, 16:57:09", "remaining_time": "14:48:51"} +{"current_steps": 5050, "total_steps": 6870, "loss": 0.2159, "lr": 9.95918992988661e-06, "epoch": 2.2052401746724892, "percentage": 73.51, "elapsed_time": "1 day, 16:59:36", "remaining_time": "14:46:25"} +{"current_steps": 5055, "total_steps": 6870, "loss": 0.2094, "lr": 9.908506279324564e-06, "epoch": 2.2074235807860263, "percentage": 73.58, "elapsed_time": "1 day, 17:02:02", "remaining_time": "14:43:59"} +{"current_steps": 5060, "total_steps": 6870, "loss": 0.2066, "lr": 9.857920031969198e-06, "epoch": 2.2096069868995634, "percentage": 73.65, "elapsed_time": "1 day, 17:04:28", "remaining_time": "14:41:33"} +{"current_steps": 5065, "total_steps": 6870, "loss": 0.2315, "lr": 9.807431514313225e-06, "epoch": 2.2117903930131004, "percentage": 73.73, "elapsed_time": "1 day, 17:06:54", "remaining_time": "14:39:07"} +{"current_steps": 5070, "total_steps": 6870, "loss": 0.2192, "lr": 9.757041052218587e-06, "epoch": 2.2139737991266375, "percentage": 73.8, "elapsed_time": "1 day, 17:09:20", "remaining_time": "14:36:41"} +{"current_steps": 5075, "total_steps": 6870, "loss": 0.21, "lr": 9.70674897091437e-06, "epoch": 2.2161572052401746, "percentage": 73.87, "elapsed_time": "1 day, 17:11:47", "remaining_time": "14:34:15"} +{"current_steps": 5080, "total_steps": 6870, "loss": 0.2025, "lr": 9.656555594994687e-06, "epoch": 2.2183406113537116, "percentage": 73.94, "elapsed_time": "1 day, 17:14:13", "remaining_time": "14:31:49"} +{"current_steps": 5085, "total_steps": 6870, "loss": 0.2016, "lr": 9.606461248416573e-06, "epoch": 2.2205240174672487, "percentage": 74.02, "elapsed_time": "1 day, 17:16:39", "remaining_time": "14:29:23"} +{"current_steps": 5090, "total_steps": 6870, "loss": 0.2098, "lr": 9.556466254497931e-06, "epoch": 2.222707423580786, "percentage": 74.09, "elapsed_time": "1 day, 17:19:05", "remaining_time": "14:26:57"} +{"current_steps": 5095, "total_steps": 6870, "loss": 0.2087, "lr": 9.506570935915419e-06, "epoch": 2.2248908296943233, "percentage": 74.16, "elapsed_time": "1 day, 17:21:32", "remaining_time": "14:24:31"} +{"current_steps": 5100, "total_steps": 6870, "loss": 0.2079, "lr": 9.456775614702364e-06, "epoch": 2.2270742358078603, "percentage": 74.24, "elapsed_time": "1 day, 17:23:58", "remaining_time": "14:22:05"} +{"current_steps": 5105, "total_steps": 6870, "loss": 0.2107, "lr": 9.407080612246697e-06, "epoch": 2.2292576419213974, "percentage": 74.31, "elapsed_time": "1 day, 17:26:24", "remaining_time": "14:19:38"} +{"current_steps": 5110, "total_steps": 6870, "loss": 0.2254, "lr": 9.357486249288891e-06, "epoch": 2.2314410480349345, "percentage": 74.38, "elapsed_time": "1 day, 17:28:50", "remaining_time": "14:17:12"} +{"current_steps": 5115, "total_steps": 6870, "loss": 0.2032, "lr": 9.307992845919844e-06, "epoch": 2.2336244541484715, "percentage": 74.45, "elapsed_time": "1 day, 17:31:16", "remaining_time": "14:14:46"} +{"current_steps": 5120, "total_steps": 6870, "loss": 0.2076, "lr": 9.258600721578872e-06, "epoch": 2.2358078602620086, "percentage": 74.53, "elapsed_time": "1 day, 17:33:42", "remaining_time": "14:12:20"} +{"current_steps": 5125, "total_steps": 6870, "loss": 0.2186, "lr": 9.209310195051582e-06, "epoch": 2.2379912663755457, "percentage": 74.6, "elapsed_time": "1 day, 17:36:09", "remaining_time": "14:09:54"} +{"current_steps": 5130, "total_steps": 6870, "loss": 0.1942, "lr": 9.160121584467878e-06, "epoch": 2.2401746724890828, "percentage": 74.67, "elapsed_time": "1 day, 17:38:35", "remaining_time": "14:07:28"} +{"current_steps": 5135, "total_steps": 6870, "loss": 0.2155, "lr": 9.111035207299878e-06, "epoch": 2.2423580786026203, "percentage": 74.75, "elapsed_time": "1 day, 17:41:01", "remaining_time": "14:05:02"} +{"current_steps": 5140, "total_steps": 6870, "loss": 0.2155, "lr": 9.062051380359857e-06, "epoch": 2.2445414847161573, "percentage": 74.82, "elapsed_time": "1 day, 17:43:27", "remaining_time": "14:02:36"} +{"current_steps": 5145, "total_steps": 6870, "loss": 0.2191, "lr": 9.01317041979822e-06, "epoch": 2.2467248908296944, "percentage": 74.89, "elapsed_time": "1 day, 17:45:53", "remaining_time": "14:00:10"} +{"current_steps": 5150, "total_steps": 6870, "loss": 0.198, "lr": 8.964392641101455e-06, "epoch": 2.2489082969432315, "percentage": 74.96, "elapsed_time": "1 day, 17:48:19", "remaining_time": "13:57:44"} +{"current_steps": 5155, "total_steps": 6870, "loss": 0.2049, "lr": 8.915718359090086e-06, "epoch": 2.2510917030567685, "percentage": 75.04, "elapsed_time": "1 day, 17:50:46", "remaining_time": "13:55:17"} +{"current_steps": 5160, "total_steps": 6870, "loss": 0.2227, "lr": 8.867147887916663e-06, "epoch": 2.2532751091703056, "percentage": 75.11, "elapsed_time": "1 day, 17:53:12", "remaining_time": "13:52:51"} +{"current_steps": 5165, "total_steps": 6870, "loss": 0.2118, "lr": 8.818681541063714e-06, "epoch": 2.2554585152838427, "percentage": 75.18, "elapsed_time": "1 day, 17:55:38", "remaining_time": "13:50:25"} +{"current_steps": 5170, "total_steps": 6870, "loss": 0.2014, "lr": 8.770319631341745e-06, "epoch": 2.2576419213973797, "percentage": 75.25, "elapsed_time": "1 day, 17:58:04", "remaining_time": "13:47:59"} +{"current_steps": 5175, "total_steps": 6870, "loss": 0.1973, "lr": 8.722062470887169e-06, "epoch": 2.2598253275109172, "percentage": 75.33, "elapsed_time": "1 day, 18:00:30", "remaining_time": "13:45:33"} +{"current_steps": 5180, "total_steps": 6870, "loss": 0.2132, "lr": 8.673910371160368e-06, "epoch": 2.2620087336244543, "percentage": 75.4, "elapsed_time": "1 day, 18:02:56", "remaining_time": "13:43:07"} +{"current_steps": 5185, "total_steps": 6870, "loss": 0.2102, "lr": 8.625863642943623e-06, "epoch": 2.2641921397379914, "percentage": 75.47, "elapsed_time": "1 day, 18:05:22", "remaining_time": "13:40:41"} +{"current_steps": 5190, "total_steps": 6870, "loss": 0.2091, "lr": 8.577922596339142e-06, "epoch": 2.2663755458515285, "percentage": 75.55, "elapsed_time": "1 day, 18:07:48", "remaining_time": "13:38:15"} +{"current_steps": 5195, "total_steps": 6870, "loss": 0.2119, "lr": 8.530087540767029e-06, "epoch": 2.2685589519650655, "percentage": 75.62, "elapsed_time": "1 day, 18:10:15", "remaining_time": "13:35:49"} +{"current_steps": 5200, "total_steps": 6870, "loss": 0.2088, "lr": 8.482358784963312e-06, "epoch": 2.2707423580786026, "percentage": 75.69, "elapsed_time": "1 day, 18:12:41", "remaining_time": "13:33:22"} +{"current_steps": 5205, "total_steps": 6870, "loss": 0.1949, "lr": 8.43473663697794e-06, "epoch": 2.2729257641921397, "percentage": 75.76, "elapsed_time": "1 day, 18:15:07", "remaining_time": "13:30:56"} +{"current_steps": 5210, "total_steps": 6870, "loss": 0.2036, "lr": 8.387221404172796e-06, "epoch": 2.2751091703056767, "percentage": 75.84, "elapsed_time": "1 day, 18:17:33", "remaining_time": "13:28:30"} +{"current_steps": 5215, "total_steps": 6870, "loss": 0.2146, "lr": 8.339813393219714e-06, "epoch": 2.277292576419214, "percentage": 75.91, "elapsed_time": "1 day, 18:19:59", "remaining_time": "13:26:04"} +{"current_steps": 5220, "total_steps": 6870, "loss": 0.2206, "lr": 8.292512910098508e-06, "epoch": 2.279475982532751, "percentage": 75.98, "elapsed_time": "1 day, 18:22:25", "remaining_time": "13:23:38"} +{"current_steps": 5225, "total_steps": 6870, "loss": 0.2199, "lr": 8.245320260094955e-06, "epoch": 2.2816593886462884, "percentage": 76.06, "elapsed_time": "1 day, 18:24:51", "remaining_time": "13:21:12"} +{"current_steps": 5230, "total_steps": 6870, "loss": 0.2168, "lr": 8.198235747798893e-06, "epoch": 2.2838427947598254, "percentage": 76.13, "elapsed_time": "1 day, 18:27:18", "remaining_time": "13:18:46"} +{"current_steps": 5235, "total_steps": 6870, "loss": 0.216, "lr": 8.151259677102211e-06, "epoch": 2.2860262008733625, "percentage": 76.2, "elapsed_time": "1 day, 18:29:44", "remaining_time": "13:16:20"} +{"current_steps": 5240, "total_steps": 6870, "loss": 0.2198, "lr": 8.104392351196885e-06, "epoch": 2.2882096069868996, "percentage": 76.27, "elapsed_time": "1 day, 18:32:10", "remaining_time": "13:13:54"} +{"current_steps": 5245, "total_steps": 6870, "loss": 0.2173, "lr": 8.057634072573048e-06, "epoch": 2.2903930131004366, "percentage": 76.35, "elapsed_time": "1 day, 18:34:36", "remaining_time": "13:11:27"} +{"current_steps": 5250, "total_steps": 6870, "loss": 0.2157, "lr": 8.010985143017008e-06, "epoch": 2.2925764192139737, "percentage": 76.42, "elapsed_time": "1 day, 18:37:02", "remaining_time": "13:09:01"} +{"current_steps": 5255, "total_steps": 6870, "loss": 0.215, "lr": 7.964445863609329e-06, "epoch": 2.2947598253275108, "percentage": 76.49, "elapsed_time": "1 day, 18:39:28", "remaining_time": "13:06:35"} +{"current_steps": 5260, "total_steps": 6870, "loss": 0.1993, "lr": 7.91801653472286e-06, "epoch": 2.2969432314410483, "percentage": 76.56, "elapsed_time": "1 day, 18:41:54", "remaining_time": "13:04:09"} +{"current_steps": 5265, "total_steps": 6870, "loss": 0.2285, "lr": 7.871697456020819e-06, "epoch": 2.2991266375545854, "percentage": 76.64, "elapsed_time": "1 day, 18:44:20", "remaining_time": "13:01:43"} +{"current_steps": 5270, "total_steps": 6870, "loss": 0.2137, "lr": 7.825488926454851e-06, "epoch": 2.3013100436681224, "percentage": 76.71, "elapsed_time": "1 day, 18:46:47", "remaining_time": "12:59:17"} +{"current_steps": 5275, "total_steps": 6870, "loss": 0.2094, "lr": 7.77939124426308e-06, "epoch": 2.3034934497816595, "percentage": 76.78, "elapsed_time": "1 day, 18:49:13", "remaining_time": "12:56:51"} +{"current_steps": 5280, "total_steps": 6870, "loss": 0.2097, "lr": 7.733404706968215e-06, "epoch": 2.3056768558951966, "percentage": 76.86, "elapsed_time": "1 day, 18:51:39", "remaining_time": "12:54:25"} +{"current_steps": 5285, "total_steps": 6870, "loss": 0.2161, "lr": 7.687529611375616e-06, "epoch": 2.3078602620087336, "percentage": 76.93, "elapsed_time": "1 day, 18:54:05", "remaining_time": "12:51:59"} +{"current_steps": 5290, "total_steps": 6870, "loss": 0.2041, "lr": 7.64176625357138e-06, "epoch": 2.3100436681222707, "percentage": 77.0, "elapsed_time": "1 day, 18:56:31", "remaining_time": "12:49:32"} +{"current_steps": 5295, "total_steps": 6870, "loss": 0.2124, "lr": 7.596114928920422e-06, "epoch": 2.3122270742358078, "percentage": 77.07, "elapsed_time": "1 day, 18:58:57", "remaining_time": "12:47:06"} +{"current_steps": 5300, "total_steps": 6870, "loss": 0.2018, "lr": 7.55057593206458e-06, "epoch": 2.314410480349345, "percentage": 77.15, "elapsed_time": "1 day, 19:01:23", "remaining_time": "12:44:40"} +{"current_steps": 5305, "total_steps": 6870, "loss": 0.2139, "lr": 7.505149556920699e-06, "epoch": 2.316593886462882, "percentage": 77.22, "elapsed_time": "1 day, 19:03:50", "remaining_time": "12:42:14"} +{"current_steps": 5310, "total_steps": 6870, "loss": 0.2153, "lr": 7.459836096678757e-06, "epoch": 2.3187772925764194, "percentage": 77.29, "elapsed_time": "1 day, 19:06:16", "remaining_time": "12:39:48"} +{"current_steps": 5315, "total_steps": 6870, "loss": 0.215, "lr": 7.41463584379995e-06, "epoch": 2.3209606986899565, "percentage": 77.37, "elapsed_time": "1 day, 19:08:42", "remaining_time": "12:37:22"} +{"current_steps": 5320, "total_steps": 6870, "loss": 0.212, "lr": 7.369549090014821e-06, "epoch": 2.3231441048034935, "percentage": 77.44, "elapsed_time": "1 day, 19:11:08", "remaining_time": "12:34:56"} +{"current_steps": 5325, "total_steps": 6870, "loss": 0.216, "lr": 7.32457612632135e-06, "epoch": 2.3253275109170306, "percentage": 77.51, "elapsed_time": "1 day, 19:13:34", "remaining_time": "12:32:30"} +{"current_steps": 5330, "total_steps": 6870, "loss": 0.2166, "lr": 7.279717242983117e-06, "epoch": 2.3275109170305677, "percentage": 77.58, "elapsed_time": "1 day, 19:16:01", "remaining_time": "12:30:04"} +{"current_steps": 5335, "total_steps": 6870, "loss": 0.207, "lr": 7.234972729527401e-06, "epoch": 2.3296943231441047, "percentage": 77.66, "elapsed_time": "1 day, 19:18:26", "remaining_time": "12:27:37"} +{"current_steps": 5340, "total_steps": 6870, "loss": 0.201, "lr": 7.190342874743311e-06, "epoch": 2.331877729257642, "percentage": 77.73, "elapsed_time": "1 day, 19:20:53", "remaining_time": "12:25:11"} +{"current_steps": 5345, "total_steps": 6870, "loss": 0.1968, "lr": 7.145827966679936e-06, "epoch": 2.334061135371179, "percentage": 77.8, "elapsed_time": "1 day, 19:23:19", "remaining_time": "12:22:45"} +{"current_steps": 5350, "total_steps": 6870, "loss": 0.2176, "lr": 7.101428292644477e-06, "epoch": 2.3362445414847164, "percentage": 77.87, "elapsed_time": "1 day, 19:25:45", "remaining_time": "12:20:19"} +{"current_steps": 5355, "total_steps": 6870, "loss": 0.2013, "lr": 7.057144139200378e-06, "epoch": 2.3384279475982535, "percentage": 77.95, "elapsed_time": "1 day, 19:28:12", "remaining_time": "12:17:53"} +{"current_steps": 5360, "total_steps": 6870, "loss": 0.1979, "lr": 7.0129757921655355e-06, "epoch": 2.3406113537117905, "percentage": 78.02, "elapsed_time": "1 day, 19:30:38", "remaining_time": "12:15:27"} +{"current_steps": 5365, "total_steps": 6870, "loss": 0.2078, "lr": 6.968923536610356e-06, "epoch": 2.3427947598253276, "percentage": 78.09, "elapsed_time": "1 day, 19:33:05", "remaining_time": "12:13:01"} +{"current_steps": 5370, "total_steps": 6870, "loss": 0.1936, "lr": 6.924987656856002e-06, "epoch": 2.3449781659388647, "percentage": 78.17, "elapsed_time": "1 day, 19:35:31", "remaining_time": "12:10:35"} +{"current_steps": 5375, "total_steps": 6870, "loss": 0.2114, "lr": 6.881168436472524e-06, "epoch": 2.3471615720524017, "percentage": 78.24, "elapsed_time": "1 day, 19:37:57", "remaining_time": "12:08:09"} +{"current_steps": 5380, "total_steps": 6870, "loss": 0.1941, "lr": 6.837466158277025e-06, "epoch": 2.349344978165939, "percentage": 78.31, "elapsed_time": "1 day, 19:40:24", "remaining_time": "12:05:43"} +{"current_steps": 5385, "total_steps": 6870, "loss": 0.2151, "lr": 6.7938811043318475e-06, "epoch": 2.351528384279476, "percentage": 78.38, "elapsed_time": "1 day, 19:42:50", "remaining_time": "12:03:17"} +{"current_steps": 5390, "total_steps": 6870, "loss": 0.2066, "lr": 6.750413555942742e-06, "epoch": 2.353711790393013, "percentage": 78.46, "elapsed_time": "1 day, 19:45:16", "remaining_time": "12:00:51"} +{"current_steps": 5395, "total_steps": 6870, "loss": 0.2168, "lr": 6.707063793657065e-06, "epoch": 2.35589519650655, "percentage": 78.53, "elapsed_time": "1 day, 19:47:42", "remaining_time": "11:58:25"} +{"current_steps": 5400, "total_steps": 6870, "loss": 0.2193, "lr": 6.663832097261952e-06, "epoch": 2.3580786026200875, "percentage": 78.6, "elapsed_time": "1 day, 19:50:08", "remaining_time": "11:55:59"} +{"current_steps": 5405, "total_steps": 6870, "loss": 0.2031, "lr": 6.620718745782525e-06, "epoch": 2.3602620087336246, "percentage": 78.68, "elapsed_time": "1 day, 19:52:35", "remaining_time": "11:53:32"} +{"current_steps": 5410, "total_steps": 6870, "loss": 0.2147, "lr": 6.577724017480094e-06, "epoch": 2.3624454148471616, "percentage": 78.75, "elapsed_time": "1 day, 19:55:01", "remaining_time": "11:51:06"} +{"current_steps": 5415, "total_steps": 6870, "loss": 0.1992, "lr": 6.534848189850329e-06, "epoch": 2.3646288209606987, "percentage": 78.82, "elapsed_time": "1 day, 19:57:27", "remaining_time": "11:48:40"} +{"current_steps": 5420, "total_steps": 6870, "loss": 0.2029, "lr": 6.492091539621517e-06, "epoch": 2.3668122270742358, "percentage": 78.89, "elapsed_time": "1 day, 19:59:53", "remaining_time": "11:46:14"} +{"current_steps": 5425, "total_steps": 6870, "loss": 0.2018, "lr": 6.449454342752748e-06, "epoch": 2.368995633187773, "percentage": 78.97, "elapsed_time": "1 day, 20:02:19", "remaining_time": "11:43:48"} +{"current_steps": 5430, "total_steps": 6870, "loss": 0.2037, "lr": 6.406936874432134e-06, "epoch": 2.37117903930131, "percentage": 79.04, "elapsed_time": "1 day, 20:04:45", "remaining_time": "11:41:22"} +{"current_steps": 5435, "total_steps": 6870, "loss": 0.2017, "lr": 6.364539409075043e-06, "epoch": 2.373362445414847, "percentage": 79.11, "elapsed_time": "1 day, 20:07:12", "remaining_time": "11:38:56"} +{"current_steps": 5440, "total_steps": 6870, "loss": 0.2093, "lr": 6.322262220322314e-06, "epoch": 2.3755458515283845, "percentage": 79.18, "elapsed_time": "1 day, 20:09:38", "remaining_time": "11:36:30"} +{"current_steps": 5445, "total_steps": 6870, "loss": 0.2031, "lr": 6.280105581038509e-06, "epoch": 2.3777292576419216, "percentage": 79.26, "elapsed_time": "1 day, 20:12:04", "remaining_time": "11:34:04"} +{"current_steps": 5450, "total_steps": 6870, "loss": 0.2082, "lr": 6.238069763310136e-06, "epoch": 2.3799126637554586, "percentage": 79.33, "elapsed_time": "1 day, 20:14:30", "remaining_time": "11:31:38"} +{"current_steps": 5455, "total_steps": 6870, "loss": 0.2066, "lr": 6.1961550384438984e-06, "epoch": 2.3820960698689957, "percentage": 79.4, "elapsed_time": "1 day, 20:16:56", "remaining_time": "11:29:11"} +{"current_steps": 5460, "total_steps": 6870, "loss": 0.2165, "lr": 6.154361676964956e-06, "epoch": 2.3842794759825328, "percentage": 79.48, "elapsed_time": "1 day, 20:19:22", "remaining_time": "11:26:45"} +{"current_steps": 5465, "total_steps": 6870, "loss": 0.2019, "lr": 6.112689948615144e-06, "epoch": 2.38646288209607, "percentage": 79.55, "elapsed_time": "1 day, 20:21:49", "remaining_time": "11:24:19"} +{"current_steps": 5470, "total_steps": 6870, "loss": 0.2018, "lr": 6.071140122351276e-06, "epoch": 2.388646288209607, "percentage": 79.62, "elapsed_time": "1 day, 20:24:15", "remaining_time": "11:21:53"} +{"current_steps": 5475, "total_steps": 6870, "loss": 0.2057, "lr": 6.029712466343382e-06, "epoch": 2.390829694323144, "percentage": 79.69, "elapsed_time": "1 day, 20:26:41", "remaining_time": "11:19:27"} +{"current_steps": 5480, "total_steps": 6870, "loss": 0.2101, "lr": 5.988407247972982e-06, "epoch": 2.393013100436681, "percentage": 79.77, "elapsed_time": "1 day, 20:29:07", "remaining_time": "11:17:01"} +{"current_steps": 5485, "total_steps": 6870, "loss": 0.214, "lr": 5.947224733831364e-06, "epoch": 2.395196506550218, "percentage": 79.84, "elapsed_time": "1 day, 20:31:33", "remaining_time": "11:14:35"} +{"current_steps": 5490, "total_steps": 6870, "loss": 0.2069, "lr": 5.906165189717855e-06, "epoch": 2.3973799126637556, "percentage": 79.91, "elapsed_time": "1 day, 20:33:59", "remaining_time": "11:12:09"} +{"current_steps": 5495, "total_steps": 6870, "loss": 0.2093, "lr": 5.8652288806381215e-06, "epoch": 2.3995633187772927, "percentage": 79.99, "elapsed_time": "1 day, 20:36:25", "remaining_time": "11:09:42"} +{"current_steps": 5500, "total_steps": 6870, "loss": 0.1989, "lr": 5.824416070802438e-06, "epoch": 2.4017467248908297, "percentage": 80.06, "elapsed_time": "1 day, 20:38:51", "remaining_time": "11:07:16"} +{"current_steps": 5505, "total_steps": 6870, "loss": 0.1988, "lr": 5.7837270236239995e-06, "epoch": 2.403930131004367, "percentage": 80.13, "elapsed_time": "1 day, 20:41:18", "remaining_time": "11:04:50"} +{"current_steps": 5510, "total_steps": 6870, "loss": 0.2042, "lr": 5.74316200171722e-06, "epoch": 2.406113537117904, "percentage": 80.2, "elapsed_time": "1 day, 20:43:44", "remaining_time": "11:02:24"} +{"current_steps": 5515, "total_steps": 6870, "loss": 0.2036, "lr": 5.70272126689601e-06, "epoch": 2.408296943231441, "percentage": 80.28, "elapsed_time": "1 day, 20:46:10", "remaining_time": "10:59:58"} +{"current_steps": 5520, "total_steps": 6870, "loss": 0.2078, "lr": 5.662405080172134e-06, "epoch": 2.410480349344978, "percentage": 80.35, "elapsed_time": "1 day, 20:48:36", "remaining_time": "10:57:32"} +{"current_steps": 5525, "total_steps": 6870, "loss": 0.2231, "lr": 5.622213701753487e-06, "epoch": 2.4126637554585155, "percentage": 80.42, "elapsed_time": "1 day, 20:51:02", "remaining_time": "10:55:06"} +{"current_steps": 5530, "total_steps": 6870, "loss": 0.2092, "lr": 5.582147391042433e-06, "epoch": 2.4148471615720526, "percentage": 80.49, "elapsed_time": "1 day, 20:53:29", "remaining_time": "10:52:40"} +{"current_steps": 5535, "total_steps": 6870, "loss": 0.2078, "lr": 5.542206406634129e-06, "epoch": 2.4170305676855897, "percentage": 80.57, "elapsed_time": "1 day, 20:55:55", "remaining_time": "10:50:14"} +{"current_steps": 5540, "total_steps": 6870, "loss": 0.2109, "lr": 5.5023910063148445e-06, "epoch": 2.4192139737991267, "percentage": 80.64, "elapsed_time": "1 day, 20:58:21", "remaining_time": "10:47:48"} +{"current_steps": 5545, "total_steps": 6870, "loss": 0.198, "lr": 5.46270144706032e-06, "epoch": 2.421397379912664, "percentage": 80.71, "elapsed_time": "1 day, 21:00:47", "remaining_time": "10:45:21"} +{"current_steps": 5550, "total_steps": 6870, "loss": 0.2138, "lr": 5.423137985034085e-06, "epoch": 2.423580786026201, "percentage": 80.79, "elapsed_time": "1 day, 21:03:13", "remaining_time": "10:42:55"} +{"current_steps": 5555, "total_steps": 6870, "loss": 0.2013, "lr": 5.3837008755858145e-06, "epoch": 2.425764192139738, "percentage": 80.86, "elapsed_time": "1 day, 21:05:39", "remaining_time": "10:40:29"} +{"current_steps": 5560, "total_steps": 6870, "loss": 0.2069, "lr": 5.3443903732496966e-06, "epoch": 2.427947598253275, "percentage": 80.93, "elapsed_time": "1 day, 21:08:05", "remaining_time": "10:38:03"} +{"current_steps": 5565, "total_steps": 6870, "loss": 0.2086, "lr": 5.305206731742748e-06, "epoch": 2.430131004366812, "percentage": 81.0, "elapsed_time": "1 day, 21:10:32", "remaining_time": "10:35:37"} +{"current_steps": 5570, "total_steps": 6870, "loss": 0.2155, "lr": 5.266150203963224e-06, "epoch": 2.432314410480349, "percentage": 81.08, "elapsed_time": "1 day, 21:12:57", "remaining_time": "10:33:11"} +{"current_steps": 5575, "total_steps": 6870, "loss": 0.2055, "lr": 5.227221041988956e-06, "epoch": 2.4344978165938866, "percentage": 81.15, "elapsed_time": "1 day, 21:15:23", "remaining_time": "10:30:45"} +{"current_steps": 5580, "total_steps": 6870, "loss": 0.2045, "lr": 5.188419497075739e-06, "epoch": 2.4366812227074237, "percentage": 81.22, "elapsed_time": "1 day, 21:17:50", "remaining_time": "10:28:19"} +{"current_steps": 5585, "total_steps": 6870, "loss": 0.2192, "lr": 5.1497458196556995e-06, "epoch": 2.438864628820961, "percentage": 81.3, "elapsed_time": "1 day, 21:20:16", "remaining_time": "10:25:52"} +{"current_steps": 5590, "total_steps": 6870, "loss": 0.2089, "lr": 5.111200259335688e-06, "epoch": 2.441048034934498, "percentage": 81.37, "elapsed_time": "1 day, 21:22:43", "remaining_time": "10:23:26"} +{"current_steps": 5595, "total_steps": 6870, "loss": 0.2243, "lr": 5.072783064895664e-06, "epoch": 2.443231441048035, "percentage": 81.44, "elapsed_time": "1 day, 21:25:10", "remaining_time": "10:21:01"} +{"current_steps": 5600, "total_steps": 6870, "loss": 0.2195, "lr": 5.034494484287086e-06, "epoch": 2.445414847161572, "percentage": 81.51, "elapsed_time": "1 day, 21:27:36", "remaining_time": "10:18:34"} +{"current_steps": 5605, "total_steps": 6870, "loss": 0.2034, "lr": 4.996334764631322e-06, "epoch": 2.447598253275109, "percentage": 81.59, "elapsed_time": "1 day, 21:30:02", "remaining_time": "10:16:08"} +{"current_steps": 5610, "total_steps": 6870, "loss": 0.2076, "lr": 4.958304152218046e-06, "epoch": 2.449781659388646, "percentage": 81.66, "elapsed_time": "1 day, 21:32:29", "remaining_time": "10:13:42"} +{"current_steps": 5615, "total_steps": 6870, "loss": 0.2142, "lr": 4.920402892503648e-06, "epoch": 2.4519650655021836, "percentage": 81.73, "elapsed_time": "1 day, 21:34:55", "remaining_time": "10:11:16"} +{"current_steps": 5620, "total_steps": 6870, "loss": 0.206, "lr": 4.882631230109655e-06, "epoch": 2.4541484716157207, "percentage": 81.8, "elapsed_time": "1 day, 21:37:21", "remaining_time": "10:08:50"} +{"current_steps": 5625, "total_steps": 6870, "loss": 0.2042, "lr": 4.844989408821149e-06, "epoch": 2.4563318777292578, "percentage": 81.88, "elapsed_time": "1 day, 21:39:47", "remaining_time": "10:06:24"} +{"current_steps": 5630, "total_steps": 6870, "loss": 0.1999, "lr": 4.807477671585195e-06, "epoch": 2.458515283842795, "percentage": 81.95, "elapsed_time": "1 day, 21:42:13", "remaining_time": "10:03:58"} +{"current_steps": 5635, "total_steps": 6870, "loss": 0.203, "lr": 4.770096260509274e-06, "epoch": 2.460698689956332, "percentage": 82.02, "elapsed_time": "1 day, 21:44:40", "remaining_time": "10:01:32"} +{"current_steps": 5640, "total_steps": 6870, "loss": 0.2134, "lr": 4.732845416859713e-06, "epoch": 2.462882096069869, "percentage": 82.1, "elapsed_time": "1 day, 21:47:06", "remaining_time": "9:59:06"} +{"current_steps": 5645, "total_steps": 6870, "loss": 0.2062, "lr": 4.695725381060137e-06, "epoch": 2.465065502183406, "percentage": 82.17, "elapsed_time": "1 day, 21:49:32", "remaining_time": "9:56:40"} +{"current_steps": 5650, "total_steps": 6870, "loss": 0.2214, "lr": 4.658736392689922e-06, "epoch": 2.467248908296943, "percentage": 82.24, "elapsed_time": "1 day, 21:51:58", "remaining_time": "9:54:13"} +{"current_steps": 5655, "total_steps": 6870, "loss": 0.2132, "lr": 4.6218786904826175e-06, "epoch": 2.46943231441048, "percentage": 82.31, "elapsed_time": "1 day, 21:54:24", "remaining_time": "9:51:47"} +{"current_steps": 5660, "total_steps": 6870, "loss": 0.2193, "lr": 4.58515251232445e-06, "epoch": 2.4716157205240172, "percentage": 82.39, "elapsed_time": "1 day, 21:56:50", "remaining_time": "9:49:21"} +{"current_steps": 5665, "total_steps": 6870, "loss": 0.206, "lr": 4.548558095252758e-06, "epoch": 2.4737991266375547, "percentage": 82.46, "elapsed_time": "1 day, 21:59:17", "remaining_time": "9:46:55"} +{"current_steps": 5670, "total_steps": 6870, "loss": 0.2116, "lr": 4.512095675454472e-06, "epoch": 2.475982532751092, "percentage": 82.53, "elapsed_time": "1 day, 22:01:43", "remaining_time": "9:44:29"} +{"current_steps": 5675, "total_steps": 6870, "loss": 0.2031, "lr": 4.475765488264588e-06, "epoch": 2.478165938864629, "percentage": 82.61, "elapsed_time": "1 day, 22:04:09", "remaining_time": "9:42:03"} +{"current_steps": 5680, "total_steps": 6870, "loss": 0.2017, "lr": 4.439567768164651e-06, "epoch": 2.480349344978166, "percentage": 82.68, "elapsed_time": "1 day, 22:06:35", "remaining_time": "9:39:37"} +{"current_steps": 5685, "total_steps": 6870, "loss": 0.2119, "lr": 4.4035027487812325e-06, "epoch": 2.482532751091703, "percentage": 82.75, "elapsed_time": "1 day, 22:09:01", "remaining_time": "9:37:11"} +{"current_steps": 5690, "total_steps": 6870, "loss": 0.1997, "lr": 4.367570662884438e-06, "epoch": 2.48471615720524, "percentage": 82.82, "elapsed_time": "1 day, 22:11:27", "remaining_time": "9:34:44"} +{"current_steps": 5695, "total_steps": 6870, "loss": 0.2096, "lr": 4.331771742386395e-06, "epoch": 2.486899563318777, "percentage": 82.9, "elapsed_time": "1 day, 22:13:53", "remaining_time": "9:32:18"} +{"current_steps": 5700, "total_steps": 6870, "loss": 0.2056, "lr": 4.296106218339757e-06, "epoch": 2.489082969432314, "percentage": 82.97, "elapsed_time": "1 day, 22:16:19", "remaining_time": "9:29:52"} +{"current_steps": 5705, "total_steps": 6870, "loss": 0.2022, "lr": 4.260574320936203e-06, "epoch": 2.4912663755458517, "percentage": 83.04, "elapsed_time": "1 day, 22:18:45", "remaining_time": "9:27:26"} +{"current_steps": 5710, "total_steps": 6870, "loss": 0.2139, "lr": 4.225176279504975e-06, "epoch": 2.493449781659389, "percentage": 83.11, "elapsed_time": "1 day, 22:21:11", "remaining_time": "9:25:00"} +{"current_steps": 5715, "total_steps": 6870, "loss": 0.2033, "lr": 4.1899123225113825e-06, "epoch": 2.495633187772926, "percentage": 83.19, "elapsed_time": "1 day, 22:23:38", "remaining_time": "9:22:34"} +{"current_steps": 5720, "total_steps": 6870, "loss": 0.2055, "lr": 4.154782677555324e-06, "epoch": 2.497816593886463, "percentage": 83.26, "elapsed_time": "1 day, 22:26:04", "remaining_time": "9:20:08"} +{"current_steps": 5725, "total_steps": 6870, "loss": 0.2056, "lr": 4.119787571369829e-06, "epoch": 2.5, "percentage": 83.33, "elapsed_time": "1 day, 22:28:30", "remaining_time": "9:17:42"} +{"current_steps": 5730, "total_steps": 6870, "loss": 0.1969, "lr": 4.084927229819591e-06, "epoch": 2.502183406113537, "percentage": 83.41, "elapsed_time": "1 day, 22:30:56", "remaining_time": "9:15:15"} +{"current_steps": 5735, "total_steps": 6870, "loss": 0.1967, "lr": 4.050201877899501e-06, "epoch": 2.504366812227074, "percentage": 83.48, "elapsed_time": "1 day, 22:33:22", "remaining_time": "9:12:49"} +{"current_steps": 5740, "total_steps": 6870, "loss": 0.2017, "lr": 4.0156117397332074e-06, "epoch": 2.506550218340611, "percentage": 83.55, "elapsed_time": "1 day, 22:35:48", "remaining_time": "9:10:23"} +{"current_steps": 5745, "total_steps": 6870, "loss": 0.2187, "lr": 3.9811570385716705e-06, "epoch": 2.5087336244541483, "percentage": 83.62, "elapsed_time": "1 day, 22:38:15", "remaining_time": "9:07:57"} +{"current_steps": 5750, "total_steps": 6870, "loss": 0.2139, "lr": 3.9468379967917025e-06, "epoch": 2.5109170305676853, "percentage": 83.7, "elapsed_time": "1 day, 22:40:41", "remaining_time": "9:05:31"} +{"current_steps": 5755, "total_steps": 6870, "loss": 0.2085, "lr": 3.912654835894564e-06, "epoch": 2.513100436681223, "percentage": 83.77, "elapsed_time": "1 day, 22:43:07", "remaining_time": "9:03:05"} +{"current_steps": 5760, "total_steps": 6870, "loss": 0.2089, "lr": 3.878607776504489e-06, "epoch": 2.51528384279476, "percentage": 83.84, "elapsed_time": "1 day, 22:45:33", "remaining_time": "9:00:39"} +{"current_steps": 5765, "total_steps": 6870, "loss": 0.2118, "lr": 3.844697038367314e-06, "epoch": 2.517467248908297, "percentage": 83.92, "elapsed_time": "1 day, 22:47:59", "remaining_time": "8:58:13"} +{"current_steps": 5770, "total_steps": 6870, "loss": 0.2041, "lr": 3.8109228403490268e-06, "epoch": 2.519650655021834, "percentage": 83.99, "elapsed_time": "1 day, 22:50:25", "remaining_time": "8:55:47"} +{"current_steps": 5775, "total_steps": 6870, "loss": 0.2117, "lr": 3.7772854004343494e-06, "epoch": 2.521834061135371, "percentage": 84.06, "elapsed_time": "1 day, 22:52:51", "remaining_time": "8:53:20"} +{"current_steps": 5780, "total_steps": 6870, "loss": 0.2051, "lr": 3.7437849357253555e-06, "epoch": 2.524017467248908, "percentage": 84.13, "elapsed_time": "1 day, 22:55:18", "remaining_time": "8:50:54"} +{"current_steps": 5785, "total_steps": 6870, "loss": 0.2112, "lr": 3.7104216624400502e-06, "epoch": 2.5262008733624453, "percentage": 84.21, "elapsed_time": "1 day, 22:57:44", "remaining_time": "8:48:28"} +{"current_steps": 5790, "total_steps": 6870, "loss": 0.2052, "lr": 3.677195795910984e-06, "epoch": 2.5283842794759828, "percentage": 84.28, "elapsed_time": "1 day, 23:00:10", "remaining_time": "8:46:02"} +{"current_steps": 5795, "total_steps": 6870, "loss": 0.2108, "lr": 3.6441075505838586e-06, "epoch": 2.53056768558952, "percentage": 84.35, "elapsed_time": "1 day, 23:02:36", "remaining_time": "8:43:36"} +{"current_steps": 5800, "total_steps": 6870, "loss": 0.2107, "lr": 3.6111571400161387e-06, "epoch": 2.532751091703057, "percentage": 84.43, "elapsed_time": "1 day, 23:05:02", "remaining_time": "8:41:10"} +{"current_steps": 5805, "total_steps": 6870, "loss": 0.2025, "lr": 3.578344776875689e-06, "epoch": 2.534934497816594, "percentage": 84.5, "elapsed_time": "1 day, 23:07:28", "remaining_time": "8:38:44"} +{"current_steps": 5810, "total_steps": 6870, "loss": 0.2089, "lr": 3.5456706729393774e-06, "epoch": 2.537117903930131, "percentage": 84.57, "elapsed_time": "1 day, 23:09:54", "remaining_time": "8:36:18"} +{"current_steps": 5815, "total_steps": 6870, "loss": 0.2136, "lr": 3.5131350390917333e-06, "epoch": 2.539301310043668, "percentage": 84.64, "elapsed_time": "1 day, 23:12:20", "remaining_time": "8:33:51"} +{"current_steps": 5820, "total_steps": 6870, "loss": 0.2206, "lr": 3.4807380853235704e-06, "epoch": 2.541484716157205, "percentage": 84.72, "elapsed_time": "1 day, 23:14:46", "remaining_time": "8:31:25"} +{"current_steps": 5825, "total_steps": 6870, "loss": 0.2104, "lr": 3.4484800207306405e-06, "epoch": 2.5436681222707422, "percentage": 84.79, "elapsed_time": "1 day, 23:17:13", "remaining_time": "8:28:59"} +{"current_steps": 5830, "total_steps": 6870, "loss": 0.2082, "lr": 3.4163610535122738e-06, "epoch": 2.5458515283842793, "percentage": 84.86, "elapsed_time": "1 day, 23:19:39", "remaining_time": "8:26:33"} +{"current_steps": 5835, "total_steps": 6870, "loss": 0.2046, "lr": 3.384381390970048e-06, "epoch": 2.5480349344978164, "percentage": 84.93, "elapsed_time": "1 day, 23:22:05", "remaining_time": "8:24:07"} +{"current_steps": 5840, "total_steps": 6870, "loss": 0.2023, "lr": 3.3525412395064375e-06, "epoch": 2.5502183406113534, "percentage": 85.01, "elapsed_time": "1 day, 23:24:31", "remaining_time": "8:21:41"} +{"current_steps": 5845, "total_steps": 6870, "loss": 0.2027, "lr": 3.32084080462349e-06, "epoch": 2.552401746724891, "percentage": 85.08, "elapsed_time": "1 day, 23:26:57", "remaining_time": "8:19:15"} +{"current_steps": 5850, "total_steps": 6870, "loss": 0.2096, "lr": 3.2892802909214955e-06, "epoch": 2.554585152838428, "percentage": 85.15, "elapsed_time": "1 day, 23:29:23", "remaining_time": "8:16:49"} +{"current_steps": 5855, "total_steps": 6870, "loss": 0.1951, "lr": 3.2578599020976707e-06, "epoch": 2.556768558951965, "percentage": 85.23, "elapsed_time": "1 day, 23:31:50", "remaining_time": "8:14:22"} +{"current_steps": 5860, "total_steps": 6870, "loss": 0.211, "lr": 3.2265798409448376e-06, "epoch": 2.558951965065502, "percentage": 85.3, "elapsed_time": "1 day, 23:34:16", "remaining_time": "8:11:56"} +{"current_steps": 5865, "total_steps": 6870, "loss": 0.2179, "lr": 3.1954403093501227e-06, "epoch": 2.561135371179039, "percentage": 85.37, "elapsed_time": "1 day, 23:36:42", "remaining_time": "8:09:30"} +{"current_steps": 5870, "total_steps": 6870, "loss": 0.2017, "lr": 3.1644415082936417e-06, "epoch": 2.5633187772925763, "percentage": 85.44, "elapsed_time": "1 day, 23:39:08", "remaining_time": "8:07:04"} +{"current_steps": 5875, "total_steps": 6870, "loss": 0.1963, "lr": 3.1335836378472233e-06, "epoch": 2.565502183406114, "percentage": 85.52, "elapsed_time": "1 day, 23:41:34", "remaining_time": "8:04:38"} +{"current_steps": 5880, "total_steps": 6870, "loss": 0.2076, "lr": 3.10286689717309e-06, "epoch": 2.567685589519651, "percentage": 85.59, "elapsed_time": "1 day, 23:44:01", "remaining_time": "8:02:12"} +{"current_steps": 5885, "total_steps": 6870, "loss": 0.211, "lr": 3.0722914845225963e-06, "epoch": 2.569868995633188, "percentage": 85.66, "elapsed_time": "1 day, 23:46:27", "remaining_time": "7:59:46"} +{"current_steps": 5890, "total_steps": 6870, "loss": 0.1983, "lr": 3.04185759723494e-06, "epoch": 2.572052401746725, "percentage": 85.74, "elapsed_time": "1 day, 23:48:53", "remaining_time": "7:57:20"} +{"current_steps": 5895, "total_steps": 6870, "loss": 0.2119, "lr": 3.0115654317358884e-06, "epoch": 2.574235807860262, "percentage": 85.81, "elapsed_time": "1 day, 23:51:19", "remaining_time": "7:54:54"} +{"current_steps": 5900, "total_steps": 6870, "loss": 0.212, "lr": 2.981415183536504e-06, "epoch": 2.576419213973799, "percentage": 85.88, "elapsed_time": "1 day, 23:53:45", "remaining_time": "7:52:27"} +{"current_steps": 5905, "total_steps": 6870, "loss": 0.1956, "lr": 2.9514070472318965e-06, "epoch": 2.578602620087336, "percentage": 85.95, "elapsed_time": "1 day, 23:56:11", "remaining_time": "7:50:01"} +{"current_steps": 5910, "total_steps": 6870, "loss": 0.2006, "lr": 2.9215412164999594e-06, "epoch": 2.5807860262008733, "percentage": 86.03, "elapsed_time": "1 day, 23:58:38", "remaining_time": "7:47:35"} +{"current_steps": 5915, "total_steps": 6870, "loss": 0.192, "lr": 2.891817884100123e-06, "epoch": 2.5829694323144103, "percentage": 86.1, "elapsed_time": "2 days, 0:01:04", "remaining_time": "7:45:09"} +{"current_steps": 5920, "total_steps": 6870, "loss": 0.2215, "lr": 2.862237241872101e-06, "epoch": 2.5851528384279474, "percentage": 86.17, "elapsed_time": "2 days, 0:03:30", "remaining_time": "7:42:43"} +{"current_steps": 5925, "total_steps": 6870, "loss": 0.2244, "lr": 2.832799480734663e-06, "epoch": 2.5873362445414845, "percentage": 86.24, "elapsed_time": "2 days, 0:05:56", "remaining_time": "7:40:17"} +{"current_steps": 5930, "total_steps": 6870, "loss": 0.2053, "lr": 2.8035047906844e-06, "epoch": 2.589519650655022, "percentage": 86.32, "elapsed_time": "2 days, 0:08:22", "remaining_time": "7:37:51"} +{"current_steps": 5935, "total_steps": 6870, "loss": 0.2047, "lr": 2.774353360794493e-06, "epoch": 2.591703056768559, "percentage": 86.39, "elapsed_time": "2 days, 0:10:48", "remaining_time": "7:35:25"} +{"current_steps": 5940, "total_steps": 6870, "loss": 0.213, "lr": 2.7453453792134986e-06, "epoch": 2.593886462882096, "percentage": 86.46, "elapsed_time": "2 days, 0:13:14", "remaining_time": "7:32:58"} +{"current_steps": 5945, "total_steps": 6870, "loss": 0.2038, "lr": 2.7164810331641366e-06, "epoch": 2.596069868995633, "percentage": 86.54, "elapsed_time": "2 days, 0:15:40", "remaining_time": "7:30:32"} +{"current_steps": 5950, "total_steps": 6870, "loss": 0.2116, "lr": 2.687760508942064e-06, "epoch": 2.5982532751091703, "percentage": 86.61, "elapsed_time": "2 days, 0:18:06", "remaining_time": "7:28:06"} +{"current_steps": 5955, "total_steps": 6870, "loss": 0.2145, "lr": 2.659183991914696e-06, "epoch": 2.6004366812227073, "percentage": 86.68, "elapsed_time": "2 days, 0:20:33", "remaining_time": "7:25:40"} +{"current_steps": 5960, "total_steps": 6870, "loss": 0.208, "lr": 2.6307516665200012e-06, "epoch": 2.6026200873362444, "percentage": 86.75, "elapsed_time": "2 days, 0:22:59", "remaining_time": "7:23:14"} +{"current_steps": 5965, "total_steps": 6870, "loss": 0.2179, "lr": 2.602463716265302e-06, "epoch": 2.604803493449782, "percentage": 86.83, "elapsed_time": "2 days, 0:25:25", "remaining_time": "7:20:48"} +{"current_steps": 5970, "total_steps": 6870, "loss": 0.2095, "lr": 2.5743203237261065e-06, "epoch": 2.606986899563319, "percentage": 86.9, "elapsed_time": "2 days, 0:27:51", "remaining_time": "7:18:22"} +{"current_steps": 5975, "total_steps": 6870, "loss": 0.2185, "lr": 2.5463216705449155e-06, "epoch": 2.609170305676856, "percentage": 86.97, "elapsed_time": "2 days, 0:30:17", "remaining_time": "7:15:56"} +{"current_steps": 5980, "total_steps": 6870, "loss": 0.2085, "lr": 2.5184679374300553e-06, "epoch": 2.611353711790393, "percentage": 87.05, "elapsed_time": "2 days, 0:32:43", "remaining_time": "7:13:30"} +{"current_steps": 5985, "total_steps": 6870, "loss": 0.2167, "lr": 2.4907593041545163e-06, "epoch": 2.61353711790393, "percentage": 87.12, "elapsed_time": "2 days, 0:35:10", "remaining_time": "7:11:03"} +{"current_steps": 5990, "total_steps": 6870, "loss": 0.1995, "lr": 2.4631959495547833e-06, "epoch": 2.6157205240174672, "percentage": 87.19, "elapsed_time": "2 days, 0:37:36", "remaining_time": "7:08:37"} +{"current_steps": 5995, "total_steps": 6870, "loss": 0.1941, "lr": 2.4357780515297e-06, "epoch": 2.6179039301310043, "percentage": 87.26, "elapsed_time": "2 days, 0:40:02", "remaining_time": "7:06:11"} +{"current_steps": 6000, "total_steps": 6870, "loss": 0.2023, "lr": 2.408505787039281e-06, "epoch": 2.6200873362445414, "percentage": 87.34, "elapsed_time": "2 days, 0:42:28", "remaining_time": "7:03:45"} +{"current_steps": 6005, "total_steps": 6870, "loss": 0.2039, "lr": 2.381379332103623e-06, "epoch": 2.6222707423580784, "percentage": 87.41, "elapsed_time": "2 days, 0:44:54", "remaining_time": "7:01:19"} +{"current_steps": 6010, "total_steps": 6870, "loss": 0.2007, "lr": 2.3543988618017237e-06, "epoch": 2.6244541484716155, "percentage": 87.48, "elapsed_time": "2 days, 0:47:20", "remaining_time": "6:58:53"} +{"current_steps": 6015, "total_steps": 6870, "loss": 0.2003, "lr": 2.327564550270381e-06, "epoch": 2.6266375545851526, "percentage": 87.55, "elapsed_time": "2 days, 0:49:46", "remaining_time": "6:56:27"} +{"current_steps": 6020, "total_steps": 6870, "loss": 0.1987, "lr": 2.3008765707030532e-06, "epoch": 2.62882096069869, "percentage": 87.63, "elapsed_time": "2 days, 0:52:12", "remaining_time": "6:54:01"} +{"current_steps": 6025, "total_steps": 6870, "loss": 0.2161, "lr": 2.2743350953487424e-06, "epoch": 2.631004366812227, "percentage": 87.7, "elapsed_time": "2 days, 0:54:38", "remaining_time": "6:51:34"} +{"current_steps": 6030, "total_steps": 6870, "loss": 0.1994, "lr": 2.2479402955108945e-06, "epoch": 2.6331877729257642, "percentage": 87.77, "elapsed_time": "2 days, 0:57:05", "remaining_time": "6:49:08"} +{"current_steps": 6035, "total_steps": 6870, "loss": 0.2071, "lr": 2.2216923415462802e-06, "epoch": 2.6353711790393013, "percentage": 87.85, "elapsed_time": "2 days, 0:59:31", "remaining_time": "6:46:42"} +{"current_steps": 6040, "total_steps": 6870, "loss": 0.1997, "lr": 2.1955914028639e-06, "epoch": 2.6375545851528384, "percentage": 87.92, "elapsed_time": "2 days, 1:01:57", "remaining_time": "6:44:16"} +{"current_steps": 6045, "total_steps": 6870, "loss": 0.2095, "lr": 2.1696376479238947e-06, "epoch": 2.6397379912663754, "percentage": 87.99, "elapsed_time": "2 days, 1:04:23", "remaining_time": "6:41:50"} +{"current_steps": 6050, "total_steps": 6870, "loss": 0.2065, "lr": 2.1438312442364467e-06, "epoch": 2.641921397379913, "percentage": 88.06, "elapsed_time": "2 days, 1:06:49", "remaining_time": "6:39:24"} +{"current_steps": 6055, "total_steps": 6870, "loss": 0.2011, "lr": 2.1181723583607177e-06, "epoch": 2.64410480349345, "percentage": 88.14, "elapsed_time": "2 days, 1:09:15", "remaining_time": "6:36:58"} +{"current_steps": 6060, "total_steps": 6870, "loss": 0.2075, "lr": 2.092661155903758e-06, "epoch": 2.646288209606987, "percentage": 88.21, "elapsed_time": "2 days, 1:11:41", "remaining_time": "6:34:32"} +{"current_steps": 6065, "total_steps": 6870, "loss": 0.1998, "lr": 2.067297801519441e-06, "epoch": 2.648471615720524, "percentage": 88.28, "elapsed_time": "2 days, 1:14:08", "remaining_time": "6:32:05"} +{"current_steps": 6070, "total_steps": 6870, "loss": 0.2066, "lr": 2.0420824589074073e-06, "epoch": 2.650655021834061, "percentage": 88.36, "elapsed_time": "2 days, 1:16:34", "remaining_time": "6:29:39"} +{"current_steps": 6075, "total_steps": 6870, "loss": 0.2078, "lr": 2.0170152908119964e-06, "epoch": 2.6528384279475983, "percentage": 88.43, "elapsed_time": "2 days, 1:19:00", "remaining_time": "6:27:13"} +{"current_steps": 6080, "total_steps": 6870, "loss": 0.2026, "lr": 1.9920964590212088e-06, "epoch": 2.6550218340611353, "percentage": 88.5, "elapsed_time": "2 days, 1:21:26", "remaining_time": "6:24:47"} +{"current_steps": 6085, "total_steps": 6870, "loss": 0.2046, "lr": 1.967326124365651e-06, "epoch": 2.6572052401746724, "percentage": 88.57, "elapsed_time": "2 days, 1:23:52", "remaining_time": "6:22:21"} +{"current_steps": 6090, "total_steps": 6870, "loss": 0.2053, "lr": 1.9427044467175054e-06, "epoch": 2.6593886462882095, "percentage": 88.65, "elapsed_time": "2 days, 1:26:18", "remaining_time": "6:19:55"} +{"current_steps": 6095, "total_steps": 6870, "loss": 0.1969, "lr": 1.9182315849894996e-06, "epoch": 2.6615720524017465, "percentage": 88.72, "elapsed_time": "2 days, 1:28:44", "remaining_time": "6:17:29"} +{"current_steps": 6100, "total_steps": 6870, "loss": 0.2133, "lr": 1.8939076971338592e-06, "epoch": 2.6637554585152836, "percentage": 88.79, "elapsed_time": "2 days, 1:31:10", "remaining_time": "6:15:03"} +{"current_steps": 6105, "total_steps": 6870, "loss": 0.2036, "lr": 1.8697329401413333e-06, "epoch": 2.665938864628821, "percentage": 88.86, "elapsed_time": "2 days, 1:33:37", "remaining_time": "6:12:36"} +{"current_steps": 6110, "total_steps": 6870, "loss": 0.2015, "lr": 1.8457074700401284e-06, "epoch": 2.668122270742358, "percentage": 88.94, "elapsed_time": "2 days, 1:36:03", "remaining_time": "6:10:10"} +{"current_steps": 6115, "total_steps": 6870, "loss": 0.2168, "lr": 1.821831441894939e-06, "epoch": 2.6703056768558953, "percentage": 89.01, "elapsed_time": "2 days, 1:38:29", "remaining_time": "6:07:44"} +{"current_steps": 6120, "total_steps": 6870, "loss": 0.2051, "lr": 1.7981050098059364e-06, "epoch": 2.6724890829694323, "percentage": 89.08, "elapsed_time": "2 days, 1:40:55", "remaining_time": "6:05:18"} +{"current_steps": 6125, "total_steps": 6870, "loss": 0.1945, "lr": 1.7745283269077622e-06, "epoch": 2.6746724890829694, "percentage": 89.16, "elapsed_time": "2 days, 1:43:21", "remaining_time": "6:02:52"} +{"current_steps": 6130, "total_steps": 6870, "loss": 0.1987, "lr": 1.751101545368558e-06, "epoch": 2.6768558951965065, "percentage": 89.23, "elapsed_time": "2 days, 1:45:47", "remaining_time": "6:00:26"} +{"current_steps": 6135, "total_steps": 6870, "loss": 0.2297, "lr": 1.7278248163889787e-06, "epoch": 2.6790393013100435, "percentage": 89.3, "elapsed_time": "2 days, 1:48:13", "remaining_time": "5:58:00"} +{"current_steps": 6140, "total_steps": 6870, "loss": 0.2004, "lr": 1.7046982902011954e-06, "epoch": 2.681222707423581, "percentage": 89.37, "elapsed_time": "2 days, 1:50:40", "remaining_time": "5:55:34"} +{"current_steps": 6145, "total_steps": 6870, "loss": 0.2165, "lr": 1.6817221160679603e-06, "epoch": 2.683406113537118, "percentage": 89.45, "elapsed_time": "2 days, 1:53:06", "remaining_time": "5:53:07"} +{"current_steps": 6150, "total_steps": 6870, "loss": 0.2139, "lr": 1.658896442281624e-06, "epoch": 2.685589519650655, "percentage": 89.52, "elapsed_time": "2 days, 1:55:32", "remaining_time": "5:50:41"} +{"current_steps": 6155, "total_steps": 6870, "loss": 0.1918, "lr": 1.636221416163175e-06, "epoch": 2.6877729257641922, "percentage": 89.59, "elapsed_time": "2 days, 1:57:58", "remaining_time": "5:48:15"} +{"current_steps": 6160, "total_steps": 6870, "loss": 0.1899, "lr": 1.6136971840612996e-06, "epoch": 2.6899563318777293, "percentage": 89.67, "elapsed_time": "2 days, 2:00:24", "remaining_time": "5:45:49"} +{"current_steps": 6165, "total_steps": 6870, "loss": 0.2034, "lr": 1.5913238913514317e-06, "epoch": 2.6921397379912664, "percentage": 89.74, "elapsed_time": "2 days, 2:02:50", "remaining_time": "5:43:23"} +{"current_steps": 6170, "total_steps": 6870, "loss": 0.2151, "lr": 1.569101682434812e-06, "epoch": 2.6943231441048034, "percentage": 89.81, "elapsed_time": "2 days, 2:05:16", "remaining_time": "5:40:57"} +{"current_steps": 6175, "total_steps": 6870, "loss": 0.1999, "lr": 1.5470307007375617e-06, "epoch": 2.6965065502183405, "percentage": 89.88, "elapsed_time": "2 days, 2:07:43", "remaining_time": "5:38:31"} +{"current_steps": 6180, "total_steps": 6870, "loss": 0.2015, "lr": 1.5251110887097547e-06, "epoch": 2.6986899563318776, "percentage": 89.96, "elapsed_time": "2 days, 2:10:09", "remaining_time": "5:36:05"} +{"current_steps": 6185, "total_steps": 6870, "loss": 0.1964, "lr": 1.5033429878245047e-06, "epoch": 2.7008733624454146, "percentage": 90.03, "elapsed_time": "2 days, 2:12:35", "remaining_time": "5:33:38"} +{"current_steps": 6190, "total_steps": 6870, "loss": 0.219, "lr": 1.4817265385770273e-06, "epoch": 2.7030567685589517, "percentage": 90.1, "elapsed_time": "2 days, 2:15:01", "remaining_time": "5:31:12"} +{"current_steps": 6195, "total_steps": 6870, "loss": 0.2087, "lr": 1.460261880483768e-06, "epoch": 2.7052401746724892, "percentage": 90.17, "elapsed_time": "2 days, 2:17:27", "remaining_time": "5:28:46"} +{"current_steps": 6200, "total_steps": 6870, "loss": 0.2073, "lr": 1.4389491520814785e-06, "epoch": 2.7074235807860263, "percentage": 90.25, "elapsed_time": "2 days, 2:19:53", "remaining_time": "5:26:20"} +{"current_steps": 6205, "total_steps": 6870, "loss": 0.2049, "lr": 1.417788490926328e-06, "epoch": 2.7096069868995634, "percentage": 90.32, "elapsed_time": "2 days, 2:22:20", "remaining_time": "5:23:54"} +{"current_steps": 6210, "total_steps": 6870, "loss": 0.1966, "lr": 1.3967800335930203e-06, "epoch": 2.7117903930131004, "percentage": 90.39, "elapsed_time": "2 days, 2:24:46", "remaining_time": "5:21:28"} +{"current_steps": 6215, "total_steps": 6870, "loss": 0.2004, "lr": 1.3759239156739046e-06, "epoch": 2.7139737991266375, "percentage": 90.47, "elapsed_time": "2 days, 2:27:12", "remaining_time": "5:19:02"} +{"current_steps": 6220, "total_steps": 6870, "loss": 0.2171, "lr": 1.355220271778107e-06, "epoch": 2.7161572052401746, "percentage": 90.54, "elapsed_time": "2 days, 2:29:38", "remaining_time": "5:16:36"} +{"current_steps": 6225, "total_steps": 6870, "loss": 0.21, "lr": 1.3346692355306583e-06, "epoch": 2.7183406113537116, "percentage": 90.61, "elapsed_time": "2 days, 2:32:04", "remaining_time": "5:14:10"} +{"current_steps": 6230, "total_steps": 6870, "loss": 0.2041, "lr": 1.3142709395716325e-06, "epoch": 2.720524017467249, "percentage": 90.68, "elapsed_time": "2 days, 2:34:30", "remaining_time": "5:11:43"} +{"current_steps": 6235, "total_steps": 6870, "loss": 0.2141, "lr": 1.294025515555289e-06, "epoch": 2.722707423580786, "percentage": 90.76, "elapsed_time": "2 days, 2:36:56", "remaining_time": "5:09:17"} +{"current_steps": 6240, "total_steps": 6870, "loss": 0.2194, "lr": 1.2739330941492211e-06, "epoch": 2.7248908296943233, "percentage": 90.83, "elapsed_time": "2 days, 2:39:23", "remaining_time": "5:06:51"} +{"current_steps": 6245, "total_steps": 6870, "loss": 0.2026, "lr": 1.253993805033521e-06, "epoch": 2.7270742358078603, "percentage": 90.9, "elapsed_time": "2 days, 2:41:49", "remaining_time": "5:04:25"} +{"current_steps": 6250, "total_steps": 6870, "loss": 0.2074, "lr": 1.2342077768999372e-06, "epoch": 2.7292576419213974, "percentage": 90.98, "elapsed_time": "2 days, 2:44:15", "remaining_time": "5:01:59"} +{"current_steps": 6255, "total_steps": 6870, "loss": 0.1982, "lr": 1.2145751374510462e-06, "epoch": 2.7314410480349345, "percentage": 91.05, "elapsed_time": "2 days, 2:46:41", "remaining_time": "4:59:33"} +{"current_steps": 6260, "total_steps": 6870, "loss": 0.2113, "lr": 1.1950960133994189e-06, "epoch": 2.7336244541484715, "percentage": 91.12, "elapsed_time": "2 days, 2:49:07", "remaining_time": "4:57:07"} +{"current_steps": 6265, "total_steps": 6870, "loss": 0.1985, "lr": 1.175770530466816e-06, "epoch": 2.7358078602620086, "percentage": 91.19, "elapsed_time": "2 days, 2:51:33", "remaining_time": "4:54:41"} +{"current_steps": 6270, "total_steps": 6870, "loss": 0.2228, "lr": 1.156598813383372e-06, "epoch": 2.7379912663755457, "percentage": 91.27, "elapsed_time": "2 days, 2:53:59", "remaining_time": "4:52:14"} +{"current_steps": 6275, "total_steps": 6870, "loss": 0.2114, "lr": 1.1375809858867853e-06, "epoch": 2.7401746724890828, "percentage": 91.34, "elapsed_time": "2 days, 2:56:25", "remaining_time": "4:49:48"} +{"current_steps": 6280, "total_steps": 6870, "loss": 0.2016, "lr": 1.1187171707215228e-06, "epoch": 2.74235807860262, "percentage": 91.41, "elapsed_time": "2 days, 2:58:51", "remaining_time": "4:47:22"} +{"current_steps": 6285, "total_steps": 6870, "loss": 0.2046, "lr": 1.1000074896380386e-06, "epoch": 2.7445414847161573, "percentage": 91.48, "elapsed_time": "2 days, 3:01:18", "remaining_time": "4:44:56"} +{"current_steps": 6290, "total_steps": 6870, "loss": 0.2063, "lr": 1.0814520633919577e-06, "epoch": 2.7467248908296944, "percentage": 91.56, "elapsed_time": "2 days, 3:03:44", "remaining_time": "4:42:30"} +{"current_steps": 6295, "total_steps": 6870, "loss": 0.2026, "lr": 1.063051011743335e-06, "epoch": 2.7489082969432315, "percentage": 91.63, "elapsed_time": "2 days, 3:06:10", "remaining_time": "4:40:04"} +{"current_steps": 6300, "total_steps": 6870, "loss": 0.2042, "lr": 1.0448044534558582e-06, "epoch": 2.7510917030567685, "percentage": 91.7, "elapsed_time": "2 days, 3:08:36", "remaining_time": "4:37:38"} +{"current_steps": 6305, "total_steps": 6870, "loss": 0.2198, "lr": 1.026712506296082e-06, "epoch": 2.7532751091703056, "percentage": 91.78, "elapsed_time": "2 days, 3:11:02", "remaining_time": "4:35:12"} +{"current_steps": 6310, "total_steps": 6870, "loss": 0.2128, "lr": 1.0087752870326817e-06, "epoch": 2.7554585152838427, "percentage": 91.85, "elapsed_time": "2 days, 3:13:28", "remaining_time": "4:32:45"} +{"current_steps": 6315, "total_steps": 6870, "loss": 0.2026, "lr": 9.909929114356842e-07, "epoch": 2.75764192139738, "percentage": 91.92, "elapsed_time": "2 days, 3:15:54", "remaining_time": "4:30:19"} +{"current_steps": 6320, "total_steps": 6870, "loss": 0.1994, "lr": 9.733654942757353e-07, "epoch": 2.7598253275109172, "percentage": 91.99, "elapsed_time": "2 days, 3:18:20", "remaining_time": "4:27:53"} +{"current_steps": 6325, "total_steps": 6870, "loss": 0.204, "lr": 9.558931493233392e-07, "epoch": 2.7620087336244543, "percentage": 92.07, "elapsed_time": "2 days, 3:20:46", "remaining_time": "4:25:27"} +{"current_steps": 6330, "total_steps": 6870, "loss": 0.1996, "lr": 9.385759893481477e-07, "epoch": 2.7641921397379914, "percentage": 92.14, "elapsed_time": "2 days, 3:23:12", "remaining_time": "4:23:01"} +{"current_steps": 6335, "total_steps": 6870, "loss": 0.2156, "lr": 9.214141261182196e-07, "epoch": 2.7663755458515285, "percentage": 92.21, "elapsed_time": "2 days, 3:25:39", "remaining_time": "4:20:35"} +{"current_steps": 6340, "total_steps": 6870, "loss": 0.208, "lr": 9.044076703992904e-07, "epoch": 2.7685589519650655, "percentage": 92.29, "elapsed_time": "2 days, 3:28:05", "remaining_time": "4:18:09"} +{"current_steps": 6345, "total_steps": 6870, "loss": 0.2074, "lr": 8.875567319540784e-07, "epoch": 2.7707423580786026, "percentage": 92.36, "elapsed_time": "2 days, 3:30:31", "remaining_time": "4:15:42"} +{"current_steps": 6350, "total_steps": 6870, "loss": 0.2, "lr": 8.708614195415604e-07, "epoch": 2.7729257641921397, "percentage": 92.43, "elapsed_time": "2 days, 3:32:57", "remaining_time": "4:13:16"} +{"current_steps": 6355, "total_steps": 6870, "loss": 0.2082, "lr": 8.543218409162778e-07, "epoch": 2.7751091703056767, "percentage": 92.5, "elapsed_time": "2 days, 3:35:23", "remaining_time": "4:10:50"} +{"current_steps": 6360, "total_steps": 6870, "loss": 0.2056, "lr": 8.379381028276401e-07, "epoch": 2.777292576419214, "percentage": 92.58, "elapsed_time": "2 days, 3:37:48", "remaining_time": "4:08:24"} +{"current_steps": 6365, "total_steps": 6870, "loss": 0.2061, "lr": 8.217103110192276e-07, "epoch": 2.779475982532751, "percentage": 92.65, "elapsed_time": "2 days, 3:40:14", "remaining_time": "4:05:58"} +{"current_steps": 6370, "total_steps": 6870, "loss": 0.2007, "lr": 8.056385702281177e-07, "epoch": 2.7816593886462884, "percentage": 92.72, "elapsed_time": "2 days, 3:42:40", "remaining_time": "4:03:32"} +{"current_steps": 6375, "total_steps": 6870, "loss": 0.2046, "lr": 7.89722984184213e-07, "epoch": 2.7838427947598254, "percentage": 92.79, "elapsed_time": "2 days, 3:45:06", "remaining_time": "4:01:06"} +{"current_steps": 6380, "total_steps": 6870, "loss": 0.2065, "lr": 7.739636556095525e-07, "epoch": 2.7860262008733625, "percentage": 92.87, "elapsed_time": "2 days, 3:47:33", "remaining_time": "3:58:40"} +{"current_steps": 6385, "total_steps": 6870, "loss": 0.2032, "lr": 7.583606862176712e-07, "epoch": 2.7882096069868996, "percentage": 92.94, "elapsed_time": "2 days, 3:49:59", "remaining_time": "3:56:13"} +{"current_steps": 6390, "total_steps": 6870, "loss": 0.2037, "lr": 7.429141767129333e-07, "epoch": 2.7903930131004366, "percentage": 93.01, "elapsed_time": "2 days, 3:52:25", "remaining_time": "3:53:47"} +{"current_steps": 6395, "total_steps": 6870, "loss": 0.2077, "lr": 7.276242267898775e-07, "epoch": 2.7925764192139737, "percentage": 93.09, "elapsed_time": "2 days, 3:54:51", "remaining_time": "3:51:21"} +{"current_steps": 6400, "total_steps": 6870, "loss": 0.2126, "lr": 7.12490935132587e-07, "epoch": 2.7947598253275108, "percentage": 93.16, "elapsed_time": "2 days, 3:57:17", "remaining_time": "3:48:55"} +{"current_steps": 6405, "total_steps": 6870, "loss": 0.2032, "lr": 6.97514399414037e-07, "epoch": 2.7969432314410483, "percentage": 93.23, "elapsed_time": "2 days, 3:59:43", "remaining_time": "3:46:29"} +{"current_steps": 6410, "total_steps": 6870, "loss": 0.2066, "lr": 6.826947162954733e-07, "epoch": 2.7991266375545854, "percentage": 93.3, "elapsed_time": "2 days, 4:02:09", "remaining_time": "3:44:03"} +{"current_steps": 6415, "total_steps": 6870, "loss": 0.2055, "lr": 6.68031981425793e-07, "epoch": 2.8013100436681224, "percentage": 93.38, "elapsed_time": "2 days, 4:04:35", "remaining_time": "3:41:37"} +{"current_steps": 6420, "total_steps": 6870, "loss": 0.2171, "lr": 6.535262894409116e-07, "epoch": 2.8034934497816595, "percentage": 93.45, "elapsed_time": "2 days, 4:07:02", "remaining_time": "3:39:11"} +{"current_steps": 6425, "total_steps": 6870, "loss": 0.1968, "lr": 6.391777339631727e-07, "epoch": 2.8056768558951966, "percentage": 93.52, "elapsed_time": "2 days, 4:09:28", "remaining_time": "3:36:44"} +{"current_steps": 6430, "total_steps": 6870, "loss": 0.2078, "lr": 6.249864076007167e-07, "epoch": 2.8078602620087336, "percentage": 93.6, "elapsed_time": "2 days, 4:11:54", "remaining_time": "3:34:18"} +{"current_steps": 6435, "total_steps": 6870, "loss": 0.2052, "lr": 6.109524019469154e-07, "epoch": 2.8100436681222707, "percentage": 93.67, "elapsed_time": "2 days, 4:14:20", "remaining_time": "3:31:52"} +{"current_steps": 6440, "total_steps": 6870, "loss": 0.1978, "lr": 5.970758075797556e-07, "epoch": 2.8122270742358078, "percentage": 93.74, "elapsed_time": "2 days, 4:16:46", "remaining_time": "3:29:26"} +{"current_steps": 6445, "total_steps": 6870, "loss": 0.205, "lr": 5.833567140612673e-07, "epoch": 2.814410480349345, "percentage": 93.81, "elapsed_time": "2 days, 4:19:13", "remaining_time": "3:27:00"} +{"current_steps": 6450, "total_steps": 6870, "loss": 0.2085, "lr": 5.69795209936938e-07, "epoch": 2.816593886462882, "percentage": 93.89, "elapsed_time": "2 days, 4:21:39", "remaining_time": "3:24:34"} +{"current_steps": 6455, "total_steps": 6870, "loss": 0.2094, "lr": 5.563913827351469e-07, "epoch": 2.818777292576419, "percentage": 93.96, "elapsed_time": "2 days, 4:24:05", "remaining_time": "3:22:08"} +{"current_steps": 6460, "total_steps": 6870, "loss": 0.2133, "lr": 5.431453189666036e-07, "epoch": 2.8209606986899565, "percentage": 94.03, "elapsed_time": "2 days, 4:26:31", "remaining_time": "3:19:42"} +{"current_steps": 6465, "total_steps": 6870, "loss": 0.1946, "lr": 5.30057104123774e-07, "epoch": 2.8231441048034935, "percentage": 94.1, "elapsed_time": "2 days, 4:28:57", "remaining_time": "3:17:15"} +{"current_steps": 6470, "total_steps": 6870, "loss": 0.2014, "lr": 5.171268226803416e-07, "epoch": 2.8253275109170306, "percentage": 94.18, "elapsed_time": "2 days, 4:31:23", "remaining_time": "3:14:49"} +{"current_steps": 6475, "total_steps": 6870, "loss": 0.2228, "lr": 5.043545580906695e-07, "epoch": 2.8275109170305677, "percentage": 94.25, "elapsed_time": "2 days, 4:33:49", "remaining_time": "3:12:23"} +{"current_steps": 6480, "total_steps": 6870, "loss": 0.208, "lr": 4.917403927892389e-07, "epoch": 2.8296943231441047, "percentage": 94.32, "elapsed_time": "2 days, 4:36:15", "remaining_time": "3:09:57"} +{"current_steps": 6485, "total_steps": 6870, "loss": 0.1994, "lr": 4.792844081901309e-07, "epoch": 2.831877729257642, "percentage": 94.4, "elapsed_time": "2 days, 4:38:41", "remaining_time": "3:07:31"} +{"current_steps": 6490, "total_steps": 6870, "loss": 0.1939, "lr": 4.6698668468651006e-07, "epoch": 2.834061135371179, "percentage": 94.47, "elapsed_time": "2 days, 4:41:07", "remaining_time": "3:05:05"} +{"current_steps": 6495, "total_steps": 6870, "loss": 0.2161, "lr": 4.548473016500854e-07, "epoch": 2.8362445414847164, "percentage": 94.54, "elapsed_time": "2 days, 4:43:34", "remaining_time": "3:02:39"} +{"current_steps": 6500, "total_steps": 6870, "loss": 0.2082, "lr": 4.4286633743061413e-07, "epoch": 2.8384279475982535, "percentage": 94.61, "elapsed_time": "2 days, 4:46:00", "remaining_time": "3:00:13"} +{"current_steps": 6505, "total_steps": 6870, "loss": 0.2066, "lr": 4.310438693553853e-07, "epoch": 2.8406113537117905, "percentage": 94.69, "elapsed_time": "2 days, 4:48:26", "remaining_time": "2:57:46"} +{"current_steps": 6510, "total_steps": 6870, "loss": 0.2093, "lr": 4.193799737287285e-07, "epoch": 2.8427947598253276, "percentage": 94.76, "elapsed_time": "2 days, 4:50:52", "remaining_time": "2:55:20"} +{"current_steps": 6515, "total_steps": 6870, "loss": 0.2235, "lr": 4.0787472583151964e-07, "epoch": 2.8449781659388647, "percentage": 94.83, "elapsed_time": "2 days, 4:53:18", "remaining_time": "2:52:54"} +{"current_steps": 6520, "total_steps": 6870, "loss": 0.1956, "lr": 3.965281999206899e-07, "epoch": 2.8471615720524017, "percentage": 94.91, "elapsed_time": "2 days, 4:55:44", "remaining_time": "2:50:28"} +{"current_steps": 6525, "total_steps": 6870, "loss": 0.2074, "lr": 3.853404692287538e-07, "epoch": 2.849344978165939, "percentage": 94.98, "elapsed_time": "2 days, 4:58:10", "remaining_time": "2:48:02"} +{"current_steps": 6530, "total_steps": 6870, "loss": 0.2174, "lr": 3.743116059633317e-07, "epoch": 2.851528384279476, "percentage": 95.05, "elapsed_time": "2 days, 5:00:37", "remaining_time": "2:45:36"} +{"current_steps": 6535, "total_steps": 6870, "loss": 0.2058, "lr": 3.6344168130668097e-07, "epoch": 2.853711790393013, "percentage": 95.12, "elapsed_time": "2 days, 5:03:03", "remaining_time": "2:43:10"} +{"current_steps": 6540, "total_steps": 6870, "loss": 0.203, "lr": 3.527307654152462e-07, "epoch": 2.85589519650655, "percentage": 95.2, "elapsed_time": "2 days, 5:05:29", "remaining_time": "2:40:44"} +{"current_steps": 6545, "total_steps": 6870, "loss": 0.189, "lr": 3.421789274192011e-07, "epoch": 2.858078602620087, "percentage": 95.27, "elapsed_time": "2 days, 5:07:55", "remaining_time": "2:38:18"} +{"current_steps": 6550, "total_steps": 6870, "loss": 0.2009, "lr": 3.3178623542199916e-07, "epoch": 2.8602620087336246, "percentage": 95.34, "elapsed_time": "2 days, 5:10:21", "remaining_time": "2:35:51"} +{"current_steps": 6555, "total_steps": 6870, "loss": 0.2125, "lr": 3.2155275649994045e-07, "epoch": 2.8624454148471616, "percentage": 95.41, "elapsed_time": "2 days, 5:12:47", "remaining_time": "2:33:25"} +{"current_steps": 6560, "total_steps": 6870, "loss": 0.2014, "lr": 3.1147855670172755e-07, "epoch": 2.8646288209606987, "percentage": 95.49, "elapsed_time": "2 days, 5:15:13", "remaining_time": "2:30:59"} +{"current_steps": 6565, "total_steps": 6870, "loss": 0.1991, "lr": 3.015637010480576e-07, "epoch": 2.8668122270742358, "percentage": 95.56, "elapsed_time": "2 days, 5:17:40", "remaining_time": "2:28:33"} +{"current_steps": 6570, "total_steps": 6870, "loss": 0.2107, "lr": 2.918082535311839e-07, "epoch": 2.868995633187773, "percentage": 95.63, "elapsed_time": "2 days, 5:20:06", "remaining_time": "2:26:07"} +{"current_steps": 6575, "total_steps": 6870, "loss": 0.1988, "lr": 2.822122771145158e-07, "epoch": 2.87117903930131, "percentage": 95.71, "elapsed_time": "2 days, 5:22:32", "remaining_time": "2:23:41"} +{"current_steps": 6580, "total_steps": 6870, "loss": 0.2006, "lr": 2.727758337322056e-07, "epoch": 2.8733624454148474, "percentage": 95.78, "elapsed_time": "2 days, 5:24:58", "remaining_time": "2:21:15"} +{"current_steps": 6585, "total_steps": 6870, "loss": 0.2107, "lr": 2.6349898428875143e-07, "epoch": 2.8755458515283845, "percentage": 95.85, "elapsed_time": "2 days, 5:27:24", "remaining_time": "2:18:49"} +{"current_steps": 6590, "total_steps": 6870, "loss": 0.2109, "lr": 2.543817886586003e-07, "epoch": 2.8777292576419216, "percentage": 95.92, "elapsed_time": "2 days, 5:29:50", "remaining_time": "2:16:22"} +{"current_steps": 6595, "total_steps": 6870, "loss": 0.2155, "lr": 2.454243056857708e-07, "epoch": 2.8799126637554586, "percentage": 96.0, "elapsed_time": "2 days, 5:32:16", "remaining_time": "2:13:56"} +{"current_steps": 6600, "total_steps": 6870, "loss": 0.2005, "lr": 2.366265931834588e-07, "epoch": 2.8820960698689957, "percentage": 96.07, "elapsed_time": "2 days, 5:34:43", "remaining_time": "2:11:30"} +{"current_steps": 6605, "total_steps": 6870, "loss": 0.2135, "lr": 2.2798870793368787e-07, "epoch": 2.8842794759825328, "percentage": 96.14, "elapsed_time": "2 days, 5:37:09", "remaining_time": "2:09:04"} +{"current_steps": 6610, "total_steps": 6870, "loss": 0.2131, "lr": 2.1951070568691222e-07, "epoch": 2.88646288209607, "percentage": 96.22, "elapsed_time": "2 days, 5:39:35", "remaining_time": "2:06:38"} +{"current_steps": 6615, "total_steps": 6870, "loss": 0.2064, "lr": 2.1119264116168104e-07, "epoch": 2.888646288209607, "percentage": 96.29, "elapsed_time": "2 days, 5:42:01", "remaining_time": "2:04:12"} +{"current_steps": 6620, "total_steps": 6870, "loss": 0.1962, "lr": 2.030345680442719e-07, "epoch": 2.890829694323144, "percentage": 96.36, "elapsed_time": "2 days, 5:44:27", "remaining_time": "2:01:46"} +{"current_steps": 6625, "total_steps": 6870, "loss": 0.2145, "lr": 1.950365389883524e-07, "epoch": 2.893013100436681, "percentage": 96.43, "elapsed_time": "2 days, 5:46:53", "remaining_time": "1:59:20"} +{"current_steps": 6630, "total_steps": 6870, "loss": 0.1986, "lr": 1.8719860561463853e-07, "epoch": 2.895196506550218, "percentage": 96.51, "elapsed_time": "2 days, 5:49:19", "remaining_time": "1:56:53"} +{"current_steps": 6635, "total_steps": 6870, "loss": 0.2087, "lr": 1.7952081851055346e-07, "epoch": 2.8973799126637556, "percentage": 96.58, "elapsed_time": "2 days, 5:51:45", "remaining_time": "1:54:27"} +{"current_steps": 6640, "total_steps": 6870, "loss": 0.208, "lr": 1.720032272299138e-07, "epoch": 2.8995633187772927, "percentage": 96.65, "elapsed_time": "2 days, 5:54:12", "remaining_time": "1:52:01"} +{"current_steps": 6645, "total_steps": 6870, "loss": 0.2159, "lr": 1.6464588029260485e-07, "epoch": 2.9017467248908297, "percentage": 96.72, "elapsed_time": "2 days, 5:56:38", "remaining_time": "1:49:35"} +{"current_steps": 6650, "total_steps": 6870, "loss": 0.2079, "lr": 1.5744882518425875e-07, "epoch": 2.903930131004367, "percentage": 96.8, "elapsed_time": "2 days, 5:59:04", "remaining_time": "1:47:09"} +{"current_steps": 6655, "total_steps": 6870, "loss": 0.2113, "lr": 1.5041210835596287e-07, "epoch": 2.906113537117904, "percentage": 96.87, "elapsed_time": "2 days, 6:01:30", "remaining_time": "1:44:43"} +{"current_steps": 6660, "total_steps": 6870, "loss": 0.2043, "lr": 1.4353577522394635e-07, "epoch": 2.908296943231441, "percentage": 96.94, "elapsed_time": "2 days, 6:03:56", "remaining_time": "1:42:17"} +{"current_steps": 6665, "total_steps": 6870, "loss": 0.2086, "lr": 1.3681987016929963e-07, "epoch": 2.910480349344978, "percentage": 97.02, "elapsed_time": "2 days, 6:06:22", "remaining_time": "1:39:51"} +{"current_steps": 6670, "total_steps": 6870, "loss": 0.2004, "lr": 1.3026443653767762e-07, "epoch": 2.9126637554585155, "percentage": 97.09, "elapsed_time": "2 days, 6:08:48", "remaining_time": "1:37:24"} +{"current_steps": 6675, "total_steps": 6870, "loss": 0.206, "lr": 1.2386951663901913e-07, "epoch": 2.9148471615720526, "percentage": 97.16, "elapsed_time": "2 days, 6:11:15", "remaining_time": "1:34:58"} +{"current_steps": 6680, "total_steps": 6870, "loss": 0.1986, "lr": 1.1763515174729179e-07, "epoch": 2.9170305676855897, "percentage": 97.23, "elapsed_time": "2 days, 6:13:41", "remaining_time": "1:32:32"} +{"current_steps": 6685, "total_steps": 6870, "loss": 0.2038, "lr": 1.115613821002004e-07, "epoch": 2.9192139737991267, "percentage": 97.31, "elapsed_time": "2 days, 6:16:07", "remaining_time": "1:30:06"} +{"current_steps": 6690, "total_steps": 6870, "loss": 0.2117, "lr": 1.0564824689894282e-07, "epoch": 2.921397379912664, "percentage": 97.38, "elapsed_time": "2 days, 6:18:33", "remaining_time": "1:27:40"} +{"current_steps": 6695, "total_steps": 6870, "loss": 0.2014, "lr": 9.989578430795732e-08, "epoch": 2.923580786026201, "percentage": 97.45, "elapsed_time": "2 days, 6:20:59", "remaining_time": "1:25:14"} +{"current_steps": 6700, "total_steps": 6870, "loss": 0.2109, "lr": 9.430403145466448e-08, "epoch": 2.925764192139738, "percentage": 97.53, "elapsed_time": "2 days, 6:23:25", "remaining_time": "1:22:48"} +{"current_steps": 6705, "total_steps": 6870, "loss": 0.2003, "lr": 8.887302442924517e-08, "epoch": 2.927947598253275, "percentage": 97.6, "elapsed_time": "2 days, 6:25:50", "remaining_time": "1:20:22"} +{"current_steps": 6710, "total_steps": 6870, "loss": 0.2094, "lr": 8.360279828438789e-08, "epoch": 2.930131004366812, "percentage": 97.67, "elapsed_time": "2 days, 6:28:16", "remaining_time": "1:17:55"} +{"current_steps": 6715, "total_steps": 6870, "loss": 0.2064, "lr": 7.849338703508069e-08, "epoch": 2.932314410480349, "percentage": 97.74, "elapsed_time": "2 days, 6:30:42", "remaining_time": "1:15:29"} +{"current_steps": 6720, "total_steps": 6870, "loss": 0.1927, "lr": 7.3544823658378e-08, "epoch": 2.934497816593886, "percentage": 97.82, "elapsed_time": "2 days, 6:33:09", "remaining_time": "1:13:03"} +{"current_steps": 6725, "total_steps": 6870, "loss": 0.2076, "lr": 6.875714009319523e-08, "epoch": 2.9366812227074237, "percentage": 97.89, "elapsed_time": "2 days, 6:35:35", "remaining_time": "1:10:37"} +{"current_steps": 6730, "total_steps": 6870, "loss": 0.2003, "lr": 6.413036724010058e-08, "epoch": 2.938864628820961, "percentage": 97.96, "elapsed_time": "2 days, 6:38:01", "remaining_time": "1:08:11"} +{"current_steps": 6735, "total_steps": 6870, "loss": 0.2144, "lr": 5.966453496111524e-08, "epoch": 2.941048034934498, "percentage": 98.03, "elapsed_time": "2 days, 6:40:27", "remaining_time": "1:05:45"} +{"current_steps": 6740, "total_steps": 6870, "loss": 0.2153, "lr": 5.5359672079524637e-08, "epoch": 2.943231441048035, "percentage": 98.11, "elapsed_time": "2 days, 6:42:53", "remaining_time": "1:03:19"} +{"current_steps": 6745, "total_steps": 6870, "loss": 0.2054, "lr": 5.121580637968137e-08, "epoch": 2.945414847161572, "percentage": 98.18, "elapsed_time": "2 days, 6:45:19", "remaining_time": "1:00:53"} +{"current_steps": 6750, "total_steps": 6870, "loss": 0.1971, "lr": 4.7232964606841454e-08, "epoch": 2.947598253275109, "percentage": 98.25, "elapsed_time": "2 days, 6:47:45", "remaining_time": "0:58:26"} +{"current_steps": 6755, "total_steps": 6870, "loss": 0.2085, "lr": 4.341117246697834e-08, "epoch": 2.9497816593886466, "percentage": 98.33, "elapsed_time": "2 days, 6:50:11", "remaining_time": "0:56:00"} +{"current_steps": 6760, "total_steps": 6870, "loss": 0.2137, "lr": 3.9750454626624745e-08, "epoch": 2.9519650655021836, "percentage": 98.4, "elapsed_time": "2 days, 6:52:37", "remaining_time": "0:53:34"} +{"current_steps": 6765, "total_steps": 6870, "loss": 0.1945, "lr": 3.6250834712708846e-08, "epoch": 2.9541484716157207, "percentage": 98.47, "elapsed_time": "2 days, 6:55:04", "remaining_time": "0:51:08"} +{"current_steps": 6770, "total_steps": 6870, "loss": 0.1865, "lr": 3.2912335312407226e-08, "epoch": 2.9563318777292578, "percentage": 98.54, "elapsed_time": "2 days, 6:57:30", "remaining_time": "0:48:42"} +{"current_steps": 6775, "total_steps": 6870, "loss": 0.2107, "lr": 2.9734977972989407e-08, "epoch": 2.958515283842795, "percentage": 98.62, "elapsed_time": "2 days, 6:59:56", "remaining_time": "0:46:16"} +{"current_steps": 6780, "total_steps": 6870, "loss": 0.2127, "lr": 2.6718783201692964e-08, "epoch": 2.960698689956332, "percentage": 98.69, "elapsed_time": "2 days, 7:02:22", "remaining_time": "0:43:50"} +{"current_steps": 6785, "total_steps": 6870, "loss": 0.2156, "lr": 2.3863770465576418e-08, "epoch": 2.962882096069869, "percentage": 98.76, "elapsed_time": "2 days, 7:04:48", "remaining_time": "0:41:24"} +{"current_steps": 6790, "total_steps": 6870, "loss": 0.2182, "lr": 2.116995819140821e-08, "epoch": 2.965065502183406, "percentage": 98.84, "elapsed_time": "2 days, 7:07:14", "remaining_time": "0:38:57"} +{"current_steps": 6795, "total_steps": 6870, "loss": 0.2009, "lr": 1.8637363765536264e-08, "epoch": 2.967248908296943, "percentage": 98.91, "elapsed_time": "2 days, 7:09:41", "remaining_time": "0:36:31"} +{"current_steps": 6800, "total_steps": 6870, "loss": 0.201, "lr": 1.6266003533774166e-08, "epoch": 2.96943231441048, "percentage": 98.98, "elapsed_time": "2 days, 7:12:07", "remaining_time": "0:34:05"} +{"current_steps": 6805, "total_steps": 6870, "loss": 0.214, "lr": 1.405589280130959e-08, "epoch": 2.9716157205240172, "percentage": 99.05, "elapsed_time": "2 days, 7:14:33", "remaining_time": "0:31:39"} +{"current_steps": 6810, "total_steps": 6870, "loss": 0.1958, "lr": 1.2007045832590492e-08, "epoch": 2.9737991266375547, "percentage": 99.13, "elapsed_time": "2 days, 7:16:59", "remaining_time": "0:29:13"} +{"current_steps": 6815, "total_steps": 6870, "loss": 0.1952, "lr": 1.0119475851247395e-08, "epoch": 2.975982532751092, "percentage": 99.2, "elapsed_time": "2 days, 7:19:25", "remaining_time": "0:26:47"} +{"current_steps": 6820, "total_steps": 6870, "loss": 0.207, "lr": 8.393195039993473e-09, "epoch": 2.978165938864629, "percentage": 99.27, "elapsed_time": "2 days, 7:21:52", "remaining_time": "0:24:21"} +{"current_steps": 6825, "total_steps": 6870, "loss": 0.2077, "lr": 6.82821454055238e-09, "epoch": 2.980349344978166, "percentage": 99.34, "elapsed_time": "2 days, 7:24:18", "remaining_time": "0:21:55"} +{"current_steps": 6830, "total_steps": 6870, "loss": 0.2, "lr": 5.42454445359164e-09, "epoch": 2.982532751091703, "percentage": 99.42, "elapsed_time": "2 days, 7:26:44", "remaining_time": "0:19:28"} +{"current_steps": 6835, "total_steps": 6870, "loss": 0.2028, "lr": 4.1821938386477075e-09, "epoch": 2.98471615720524, "percentage": 99.49, "elapsed_time": "2 days, 7:29:10", "remaining_time": "0:17:02"} +{"current_steps": 6840, "total_steps": 6870, "loss": 0.2119, "lr": 3.1011707140732317e-09, "epoch": 2.986899563318777, "percentage": 99.56, "elapsed_time": "2 days, 7:31:36", "remaining_time": "0:14:36"} +{"current_steps": 6845, "total_steps": 6870, "loss": 0.2146, "lr": 2.1814820569898698e-09, "epoch": 2.9890829694323147, "percentage": 99.64, "elapsed_time": "2 days, 7:34:02", "remaining_time": "0:12:10"} +{"current_steps": 6850, "total_steps": 6870, "loss": 0.213, "lr": 1.4231338032300034e-09, "epoch": 2.9912663755458517, "percentage": 99.71, "elapsed_time": "2 days, 7:36:29", "remaining_time": "0:09:44"} +{"current_steps": 6855, "total_steps": 6870, "loss": 0.2045, "lr": 8.261308473089813e-10, "epoch": 2.993449781659389, "percentage": 99.78, "elapsed_time": "2 days, 7:38:55", "remaining_time": "0:07:18"} +{"current_steps": 6860, "total_steps": 6870, "loss": 0.2067, "lr": 3.9047704239181247e-10, "epoch": 2.995633187772926, "percentage": 99.85, "elapsed_time": "2 days, 7:41:21", "remaining_time": "0:04:52"} +{"current_steps": 6865, "total_steps": 6870, "loss": 0.1943, "lr": 1.1617520026818706e-10, "epoch": 2.997816593886463, "percentage": 99.93, "elapsed_time": "2 days, 7:43:47", "remaining_time": "0:02:26"} +{"current_steps": 6870, "total_steps": 6870, "loss": 0.1971, "lr": 3.2270913274956573e-12, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2 days, 7:46:13", "remaining_time": "0:00:00"} +{"current_steps": 6870, "total_steps": 6870, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2 days, 7:46:52", "remaining_time": "0:00:00"} diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..5fd033b --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,9661 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 6870, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002183406113537118, + "grad_norm": 15.530065061351076, + "learning_rate": 2.9112081513828244e-07, + "loss": 3.5144, + "step": 5 + }, + { + "epoch": 0.004366812227074236, + "grad_norm": 14.048089091944986, + "learning_rate": 6.550218340611353e-07, + "loss": 3.3991, + "step": 10 + }, + { + "epoch": 0.006550218340611353, + "grad_norm": 10.38275753075565, + "learning_rate": 1.0189228529839884e-06, + "loss": 3.4492, + "step": 15 + }, + { + "epoch": 0.008733624454148471, + "grad_norm": 9.374594765056237, + "learning_rate": 1.3828238719068415e-06, + "loss": 3.3321, + "step": 20 + }, + { + "epoch": 0.010917030567685589, + "grad_norm": 6.58125566549559, + "learning_rate": 1.7467248908296944e-06, + "loss": 3.0597, + "step": 25 + }, + { + "epoch": 0.013100436681222707, + "grad_norm": 7.80412598343367, + "learning_rate": 2.1106259097525473e-06, + "loss": 2.9508, + "step": 30 + }, + { + "epoch": 0.015283842794759825, + "grad_norm": 7.524557299709741, + "learning_rate": 2.4745269286754004e-06, + "loss": 2.612, + "step": 35 + }, + { + "epoch": 0.017467248908296942, + "grad_norm": 6.2330781402437365, + "learning_rate": 2.8384279475982535e-06, + "loss": 2.3435, + "step": 40 + }, + { + "epoch": 0.019650655021834062, + "grad_norm": 5.3971059460100825, + "learning_rate": 3.2023289665211066e-06, + "loss": 1.9812, + "step": 45 + }, + { + "epoch": 0.021834061135371178, + "grad_norm": 5.173910417939837, + "learning_rate": 3.5662299854439593e-06, + "loss": 1.6557, + "step": 50 + }, + { + "epoch": 0.024017467248908297, + "grad_norm": 4.343854825431366, + "learning_rate": 3.930131004366813e-06, + "loss": 1.4334, + "step": 55 + }, + { + "epoch": 0.026200873362445413, + "grad_norm": 3.318343002698737, + "learning_rate": 4.2940320232896655e-06, + "loss": 1.2603, + "step": 60 + }, + { + "epoch": 0.028384279475982533, + "grad_norm": 3.3649234822213425, + "learning_rate": 4.657933042212519e-06, + "loss": 1.128, + "step": 65 + }, + { + "epoch": 0.03056768558951965, + "grad_norm": 4.651497105373944, + "learning_rate": 5.021834061135372e-06, + "loss": 1.0685, + "step": 70 + }, + { + "epoch": 0.03275109170305677, + "grad_norm": 2.1122195307286127, + "learning_rate": 5.385735080058224e-06, + "loss": 0.9865, + "step": 75 + }, + { + "epoch": 0.034934497816593885, + "grad_norm": 3.264835592266701, + "learning_rate": 5.749636098981077e-06, + "loss": 0.8563, + "step": 80 + }, + { + "epoch": 0.03711790393013101, + "grad_norm": 4.854625880384626, + "learning_rate": 6.11353711790393e-06, + "loss": 0.813, + "step": 85 + }, + { + "epoch": 0.039301310043668124, + "grad_norm": 2.2913931555435223, + "learning_rate": 6.477438136826783e-06, + "loss": 0.8189, + "step": 90 + }, + { + "epoch": 0.04148471615720524, + "grad_norm": 1.4161513553023397, + "learning_rate": 6.841339155749636e-06, + "loss": 0.7341, + "step": 95 + }, + { + "epoch": 0.043668122270742356, + "grad_norm": 1.2517988600799463, + "learning_rate": 7.205240174672489e-06, + "loss": 0.7264, + "step": 100 + }, + { + "epoch": 0.04585152838427948, + "grad_norm": 4.403128247316299, + "learning_rate": 7.569141193595343e-06, + "loss": 0.7105, + "step": 105 + }, + { + "epoch": 0.048034934497816595, + "grad_norm": 2.6065584518468996, + "learning_rate": 7.933042212518195e-06, + "loss": 0.6883, + "step": 110 + }, + { + "epoch": 0.05021834061135371, + "grad_norm": 1.2548695388960658, + "learning_rate": 8.296943231441049e-06, + "loss": 0.6444, + "step": 115 + }, + { + "epoch": 0.05240174672489083, + "grad_norm": 1.2420608105158033, + "learning_rate": 8.660844250363902e-06, + "loss": 0.6228, + "step": 120 + }, + { + "epoch": 0.05458515283842795, + "grad_norm": 1.6213174255897973, + "learning_rate": 9.024745269286754e-06, + "loss": 0.6091, + "step": 125 + }, + { + "epoch": 0.056768558951965066, + "grad_norm": 1.1139361754005903, + "learning_rate": 9.388646288209607e-06, + "loss": 0.6049, + "step": 130 + }, + { + "epoch": 0.05895196506550218, + "grad_norm": 2.019221687744237, + "learning_rate": 9.75254730713246e-06, + "loss": 0.6334, + "step": 135 + }, + { + "epoch": 0.0611353711790393, + "grad_norm": 1.4614530692241856, + "learning_rate": 1.0116448326055314e-05, + "loss": 0.5484, + "step": 140 + }, + { + "epoch": 0.06331877729257641, + "grad_norm": 1.7546701255165822, + "learning_rate": 1.0480349344978165e-05, + "loss": 0.5777, + "step": 145 + }, + { + "epoch": 0.06550218340611354, + "grad_norm": 1.5028564001758502, + "learning_rate": 1.084425036390102e-05, + "loss": 0.5427, + "step": 150 + }, + { + "epoch": 0.06768558951965066, + "grad_norm": 0.8737189391813837, + "learning_rate": 1.1208151382823872e-05, + "loss": 0.5501, + "step": 155 + }, + { + "epoch": 0.06986899563318777, + "grad_norm": 0.7913219748536747, + "learning_rate": 1.1572052401746727e-05, + "loss": 0.5404, + "step": 160 + }, + { + "epoch": 0.07205240174672489, + "grad_norm": 1.2931917942966242, + "learning_rate": 1.1935953420669578e-05, + "loss": 0.5317, + "step": 165 + }, + { + "epoch": 0.07423580786026202, + "grad_norm": 1.7815089914348687, + "learning_rate": 1.2299854439592432e-05, + "loss": 0.5391, + "step": 170 + }, + { + "epoch": 0.07641921397379912, + "grad_norm": 1.588354964609756, + "learning_rate": 1.2663755458515283e-05, + "loss": 0.5163, + "step": 175 + }, + { + "epoch": 0.07860262008733625, + "grad_norm": 0.7324777658190085, + "learning_rate": 1.3027656477438139e-05, + "loss": 0.5206, + "step": 180 + }, + { + "epoch": 0.08078602620087336, + "grad_norm": 0.9320526641008475, + "learning_rate": 1.339155749636099e-05, + "loss": 0.51, + "step": 185 + }, + { + "epoch": 0.08296943231441048, + "grad_norm": 1.0473996131197105, + "learning_rate": 1.3755458515283843e-05, + "loss": 0.4995, + "step": 190 + }, + { + "epoch": 0.0851528384279476, + "grad_norm": 1.1005598187154526, + "learning_rate": 1.4119359534206697e-05, + "loss": 0.4992, + "step": 195 + }, + { + "epoch": 0.08733624454148471, + "grad_norm": 1.3835272450195761, + "learning_rate": 1.448326055312955e-05, + "loss": 0.5157, + "step": 200 + }, + { + "epoch": 0.08951965065502183, + "grad_norm": 0.8363321954594061, + "learning_rate": 1.4847161572052404e-05, + "loss": 0.4884, + "step": 205 + }, + { + "epoch": 0.09170305676855896, + "grad_norm": 0.6751320611923934, + "learning_rate": 1.5211062590975255e-05, + "loss": 0.4837, + "step": 210 + }, + { + "epoch": 0.09388646288209607, + "grad_norm": 0.6597124878200359, + "learning_rate": 1.557496360989811e-05, + "loss": 0.4614, + "step": 215 + }, + { + "epoch": 0.09606986899563319, + "grad_norm": 1.0002616519579557, + "learning_rate": 1.5938864628820962e-05, + "loss": 0.5178, + "step": 220 + }, + { + "epoch": 0.0982532751091703, + "grad_norm": 2.094018394615991, + "learning_rate": 1.6302765647743815e-05, + "loss": 0.4785, + "step": 225 + }, + { + "epoch": 0.10043668122270742, + "grad_norm": 0.8762721338958381, + "learning_rate": 1.6666666666666667e-05, + "loss": 0.4867, + "step": 230 + }, + { + "epoch": 0.10262008733624454, + "grad_norm": 1.3388349700796345, + "learning_rate": 1.703056768558952e-05, + "loss": 0.4781, + "step": 235 + }, + { + "epoch": 0.10480349344978165, + "grad_norm": 1.0507336135636423, + "learning_rate": 1.7394468704512373e-05, + "loss": 0.4732, + "step": 240 + }, + { + "epoch": 0.10698689956331878, + "grad_norm": 0.7938940927456775, + "learning_rate": 1.7758369723435225e-05, + "loss": 0.4651, + "step": 245 + }, + { + "epoch": 0.1091703056768559, + "grad_norm": 1.0437122295269432, + "learning_rate": 1.812227074235808e-05, + "loss": 0.4585, + "step": 250 + }, + { + "epoch": 0.11135371179039301, + "grad_norm": 0.67350165285708, + "learning_rate": 1.848617176128093e-05, + "loss": 0.4609, + "step": 255 + }, + { + "epoch": 0.11353711790393013, + "grad_norm": 0.5808251314496068, + "learning_rate": 1.8850072780203784e-05, + "loss": 0.4549, + "step": 260 + }, + { + "epoch": 0.11572052401746726, + "grad_norm": 1.8396131242666172, + "learning_rate": 1.921397379912664e-05, + "loss": 0.4783, + "step": 265 + }, + { + "epoch": 0.11790393013100436, + "grad_norm": 0.9680443797016299, + "learning_rate": 1.9577874818049492e-05, + "loss": 0.455, + "step": 270 + }, + { + "epoch": 0.12008733624454149, + "grad_norm": 0.7014587496843943, + "learning_rate": 1.9941775836972345e-05, + "loss": 0.4479, + "step": 275 + }, + { + "epoch": 0.1222707423580786, + "grad_norm": 0.9218263081064745, + "learning_rate": 2.0305676855895198e-05, + "loss": 0.456, + "step": 280 + }, + { + "epoch": 0.12445414847161572, + "grad_norm": 0.5840696636129238, + "learning_rate": 2.066957787481805e-05, + "loss": 0.4318, + "step": 285 + }, + { + "epoch": 0.12663755458515283, + "grad_norm": 0.6901911501394837, + "learning_rate": 2.1033478893740903e-05, + "loss": 0.4494, + "step": 290 + }, + { + "epoch": 0.12882096069868995, + "grad_norm": 0.5256416173639158, + "learning_rate": 2.1397379912663756e-05, + "loss": 0.4604, + "step": 295 + }, + { + "epoch": 0.13100436681222707, + "grad_norm": 0.8643348227959773, + "learning_rate": 2.176128093158661e-05, + "loss": 0.4529, + "step": 300 + }, + { + "epoch": 0.1331877729257642, + "grad_norm": 0.845191908182343, + "learning_rate": 2.2125181950509464e-05, + "loss": 0.4462, + "step": 305 + }, + { + "epoch": 0.13537117903930132, + "grad_norm": 0.6911667544775351, + "learning_rate": 2.2489082969432314e-05, + "loss": 0.4353, + "step": 310 + }, + { + "epoch": 0.13755458515283842, + "grad_norm": 0.6842730240845932, + "learning_rate": 2.2852983988355166e-05, + "loss": 0.4425, + "step": 315 + }, + { + "epoch": 0.13973799126637554, + "grad_norm": 0.6095243486936803, + "learning_rate": 2.3216885007278022e-05, + "loss": 0.4375, + "step": 320 + }, + { + "epoch": 0.14192139737991266, + "grad_norm": 0.9176516550209551, + "learning_rate": 2.3580786026200875e-05, + "loss": 0.4539, + "step": 325 + }, + { + "epoch": 0.14410480349344978, + "grad_norm": 0.8430335294570548, + "learning_rate": 2.3944687045123728e-05, + "loss": 0.4347, + "step": 330 + }, + { + "epoch": 0.1462882096069869, + "grad_norm": 0.7088826431924614, + "learning_rate": 2.430858806404658e-05, + "loss": 0.4149, + "step": 335 + }, + { + "epoch": 0.14847161572052403, + "grad_norm": 0.6180099894880557, + "learning_rate": 2.4672489082969433e-05, + "loss": 0.422, + "step": 340 + }, + { + "epoch": 0.15065502183406113, + "grad_norm": 0.6817744275567881, + "learning_rate": 2.503639010189229e-05, + "loss": 0.4418, + "step": 345 + }, + { + "epoch": 0.15283842794759825, + "grad_norm": 0.6940745586404697, + "learning_rate": 2.540029112081514e-05, + "loss": 0.4404, + "step": 350 + }, + { + "epoch": 0.15502183406113537, + "grad_norm": 0.5754528111989647, + "learning_rate": 2.576419213973799e-05, + "loss": 0.4263, + "step": 355 + }, + { + "epoch": 0.1572052401746725, + "grad_norm": 1.0591998260395188, + "learning_rate": 2.6128093158660844e-05, + "loss": 0.4299, + "step": 360 + }, + { + "epoch": 0.15938864628820962, + "grad_norm": 0.7123628186533127, + "learning_rate": 2.64919941775837e-05, + "loss": 0.4196, + "step": 365 + }, + { + "epoch": 0.1615720524017467, + "grad_norm": 0.6457647967870623, + "learning_rate": 2.6855895196506553e-05, + "loss": 0.4174, + "step": 370 + }, + { + "epoch": 0.16375545851528384, + "grad_norm": 0.9935438036526899, + "learning_rate": 2.7219796215429405e-05, + "loss": 0.3993, + "step": 375 + }, + { + "epoch": 0.16593886462882096, + "grad_norm": 0.883212399644436, + "learning_rate": 2.7583697234352258e-05, + "loss": 0.4345, + "step": 380 + }, + { + "epoch": 0.16812227074235808, + "grad_norm": 0.948205577828456, + "learning_rate": 2.794759825327511e-05, + "loss": 0.4441, + "step": 385 + }, + { + "epoch": 0.1703056768558952, + "grad_norm": 0.8223542412825284, + "learning_rate": 2.831149927219796e-05, + "loss": 0.4029, + "step": 390 + }, + { + "epoch": 0.17248908296943233, + "grad_norm": 0.7182654852607071, + "learning_rate": 2.867540029112082e-05, + "loss": 0.4342, + "step": 395 + }, + { + "epoch": 0.17467248908296942, + "grad_norm": 0.6862331015738732, + "learning_rate": 2.9039301310043672e-05, + "loss": 0.4044, + "step": 400 + }, + { + "epoch": 0.17685589519650655, + "grad_norm": 0.9719904263631477, + "learning_rate": 2.940320232896652e-05, + "loss": 0.4114, + "step": 405 + }, + { + "epoch": 0.17903930131004367, + "grad_norm": 0.8428735007204966, + "learning_rate": 2.9767103347889374e-05, + "loss": 0.4124, + "step": 410 + }, + { + "epoch": 0.1812227074235808, + "grad_norm": 0.8900932432816406, + "learning_rate": 3.0131004366812227e-05, + "loss": 0.4096, + "step": 415 + }, + { + "epoch": 0.18340611353711792, + "grad_norm": 0.658732643364676, + "learning_rate": 3.0494905385735083e-05, + "loss": 0.4019, + "step": 420 + }, + { + "epoch": 0.185589519650655, + "grad_norm": 0.5717380212732919, + "learning_rate": 3.0858806404657935e-05, + "loss": 0.4125, + "step": 425 + }, + { + "epoch": 0.18777292576419213, + "grad_norm": 0.7561360040926198, + "learning_rate": 3.122270742358079e-05, + "loss": 0.3979, + "step": 430 + }, + { + "epoch": 0.18995633187772926, + "grad_norm": 0.7154647354779691, + "learning_rate": 3.158660844250364e-05, + "loss": 0.411, + "step": 435 + }, + { + "epoch": 0.19213973799126638, + "grad_norm": 1.106580393251376, + "learning_rate": 3.1950509461426493e-05, + "loss": 0.4001, + "step": 440 + }, + { + "epoch": 0.1943231441048035, + "grad_norm": 0.8277158276869705, + "learning_rate": 3.2314410480349346e-05, + "loss": 0.3982, + "step": 445 + }, + { + "epoch": 0.1965065502183406, + "grad_norm": 0.6695298962156193, + "learning_rate": 3.26783114992722e-05, + "loss": 0.4129, + "step": 450 + }, + { + "epoch": 0.19868995633187772, + "grad_norm": 0.8926693328948101, + "learning_rate": 3.304221251819505e-05, + "loss": 0.4173, + "step": 455 + }, + { + "epoch": 0.20087336244541484, + "grad_norm": 0.9134165249644083, + "learning_rate": 3.3406113537117904e-05, + "loss": 0.3947, + "step": 460 + }, + { + "epoch": 0.20305676855895197, + "grad_norm": 0.6780077341796262, + "learning_rate": 3.377001455604076e-05, + "loss": 0.4066, + "step": 465 + }, + { + "epoch": 0.2052401746724891, + "grad_norm": 0.8359819073368681, + "learning_rate": 3.413391557496361e-05, + "loss": 0.4054, + "step": 470 + }, + { + "epoch": 0.2074235807860262, + "grad_norm": 0.49801067958580947, + "learning_rate": 3.449781659388647e-05, + "loss": 0.3923, + "step": 475 + }, + { + "epoch": 0.2096069868995633, + "grad_norm": 0.7389933814419388, + "learning_rate": 3.486171761280932e-05, + "loss": 0.4068, + "step": 480 + }, + { + "epoch": 0.21179039301310043, + "grad_norm": 0.4893296535673842, + "learning_rate": 3.5225618631732174e-05, + "loss": 0.4002, + "step": 485 + }, + { + "epoch": 0.21397379912663755, + "grad_norm": 0.5093678162190949, + "learning_rate": 3.558951965065502e-05, + "loss": 0.4068, + "step": 490 + }, + { + "epoch": 0.21615720524017468, + "grad_norm": 0.9016225376861969, + "learning_rate": 3.595342066957787e-05, + "loss": 0.4081, + "step": 495 + }, + { + "epoch": 0.2183406113537118, + "grad_norm": 0.5844526927479975, + "learning_rate": 3.6317321688500726e-05, + "loss": 0.4053, + "step": 500 + }, + { + "epoch": 0.2205240174672489, + "grad_norm": 0.6328384104413857, + "learning_rate": 3.6681222707423585e-05, + "loss": 0.391, + "step": 505 + }, + { + "epoch": 0.22270742358078602, + "grad_norm": 0.6295475420727077, + "learning_rate": 3.704512372634644e-05, + "loss": 0.4061, + "step": 510 + }, + { + "epoch": 0.22489082969432314, + "grad_norm": 0.6065260581614148, + "learning_rate": 3.740902474526929e-05, + "loss": 0.4107, + "step": 515 + }, + { + "epoch": 0.22707423580786026, + "grad_norm": 1.004727626897767, + "learning_rate": 3.777292576419214e-05, + "loss": 0.3998, + "step": 520 + }, + { + "epoch": 0.2292576419213974, + "grad_norm": 1.1129157537413497, + "learning_rate": 3.8136826783114996e-05, + "loss": 0.4005, + "step": 525 + }, + { + "epoch": 0.2314410480349345, + "grad_norm": 0.9960094773975019, + "learning_rate": 3.850072780203784e-05, + "loss": 0.4098, + "step": 530 + }, + { + "epoch": 0.2336244541484716, + "grad_norm": 0.5262503731188081, + "learning_rate": 3.88646288209607e-05, + "loss": 0.3965, + "step": 535 + }, + { + "epoch": 0.23580786026200873, + "grad_norm": 0.6294913749088836, + "learning_rate": 3.9228529839883554e-05, + "loss": 0.3962, + "step": 540 + }, + { + "epoch": 0.23799126637554585, + "grad_norm": 0.6672651864348986, + "learning_rate": 3.9592430858806406e-05, + "loss": 0.3976, + "step": 545 + }, + { + "epoch": 0.24017467248908297, + "grad_norm": 0.6736374930092932, + "learning_rate": 3.995633187772926e-05, + "loss": 0.3932, + "step": 550 + }, + { + "epoch": 0.2423580786026201, + "grad_norm": 0.9025467677457151, + "learning_rate": 4.032023289665211e-05, + "loss": 0.4163, + "step": 555 + }, + { + "epoch": 0.2445414847161572, + "grad_norm": 0.6615595441983655, + "learning_rate": 4.068413391557497e-05, + "loss": 0.3885, + "step": 560 + }, + { + "epoch": 0.24672489082969432, + "grad_norm": 0.747795749818688, + "learning_rate": 4.104803493449782e-05, + "loss": 0.3979, + "step": 565 + }, + { + "epoch": 0.24890829694323144, + "grad_norm": 0.5395748717140287, + "learning_rate": 4.141193595342067e-05, + "loss": 0.4052, + "step": 570 + }, + { + "epoch": 0.25109170305676853, + "grad_norm": 0.5834684407168078, + "learning_rate": 4.177583697234352e-05, + "loss": 0.3931, + "step": 575 + }, + { + "epoch": 0.25327510917030566, + "grad_norm": 0.6149399366377413, + "learning_rate": 4.2139737991266375e-05, + "loss": 0.4199, + "step": 580 + }, + { + "epoch": 0.2554585152838428, + "grad_norm": 0.6439187925066173, + "learning_rate": 4.250363901018923e-05, + "loss": 0.3817, + "step": 585 + }, + { + "epoch": 0.2576419213973799, + "grad_norm": 0.7335269690355521, + "learning_rate": 4.286754002911209e-05, + "loss": 0.4146, + "step": 590 + }, + { + "epoch": 0.259825327510917, + "grad_norm": 0.5948480037959886, + "learning_rate": 4.323144104803494e-05, + "loss": 0.3894, + "step": 595 + }, + { + "epoch": 0.26200873362445415, + "grad_norm": 0.7265886088550407, + "learning_rate": 4.359534206695779e-05, + "loss": 0.4099, + "step": 600 + }, + { + "epoch": 0.26419213973799127, + "grad_norm": 0.8146189489413153, + "learning_rate": 4.395924308588064e-05, + "loss": 0.3937, + "step": 605 + }, + { + "epoch": 0.2663755458515284, + "grad_norm": 0.7479919188399701, + "learning_rate": 4.432314410480349e-05, + "loss": 0.4041, + "step": 610 + }, + { + "epoch": 0.2685589519650655, + "grad_norm": 0.6451793400170305, + "learning_rate": 4.468704512372635e-05, + "loss": 0.3923, + "step": 615 + }, + { + "epoch": 0.27074235807860264, + "grad_norm": 0.6731070339864949, + "learning_rate": 4.50509461426492e-05, + "loss": 0.3933, + "step": 620 + }, + { + "epoch": 0.27292576419213976, + "grad_norm": 0.7633021827753019, + "learning_rate": 4.5414847161572056e-05, + "loss": 0.3887, + "step": 625 + }, + { + "epoch": 0.27510917030567683, + "grad_norm": 0.5318275158910994, + "learning_rate": 4.577874818049491e-05, + "loss": 0.3767, + "step": 630 + }, + { + "epoch": 0.27729257641921395, + "grad_norm": 0.6053347759447276, + "learning_rate": 4.614264919941776e-05, + "loss": 0.3996, + "step": 635 + }, + { + "epoch": 0.2794759825327511, + "grad_norm": 0.8239318519157607, + "learning_rate": 4.6506550218340614e-05, + "loss": 0.3966, + "step": 640 + }, + { + "epoch": 0.2816593886462882, + "grad_norm": 0.6239015579696312, + "learning_rate": 4.687045123726347e-05, + "loss": 0.4006, + "step": 645 + }, + { + "epoch": 0.2838427947598253, + "grad_norm": 0.5211608695178815, + "learning_rate": 4.723435225618632e-05, + "loss": 0.3923, + "step": 650 + }, + { + "epoch": 0.28602620087336245, + "grad_norm": 0.6758177772158337, + "learning_rate": 4.759825327510917e-05, + "loss": 0.402, + "step": 655 + }, + { + "epoch": 0.28820960698689957, + "grad_norm": 0.7053141851010356, + "learning_rate": 4.7962154294032025e-05, + "loss": 0.4099, + "step": 660 + }, + { + "epoch": 0.2903930131004367, + "grad_norm": 0.6398424235305709, + "learning_rate": 4.832605531295488e-05, + "loss": 0.3886, + "step": 665 + }, + { + "epoch": 0.2925764192139738, + "grad_norm": 0.6225872999277295, + "learning_rate": 4.868995633187774e-05, + "loss": 0.3706, + "step": 670 + }, + { + "epoch": 0.29475982532751094, + "grad_norm": 0.5559572514117758, + "learning_rate": 4.905385735080058e-05, + "loss": 0.4012, + "step": 675 + }, + { + "epoch": 0.29694323144104806, + "grad_norm": 0.5432885290294175, + "learning_rate": 4.9417758369723435e-05, + "loss": 0.3842, + "step": 680 + }, + { + "epoch": 0.29912663755458513, + "grad_norm": 0.6340294335348677, + "learning_rate": 4.978165938864629e-05, + "loss": 0.3855, + "step": 685 + }, + { + "epoch": 0.30131004366812225, + "grad_norm": 0.7186273990805551, + "learning_rate": 4.999998709163553e-05, + "loss": 0.3705, + "step": 690 + }, + { + "epoch": 0.3034934497816594, + "grad_norm": 0.6162542024063481, + "learning_rate": 4.999984187268831e-05, + "loss": 0.3828, + "step": 695 + }, + { + "epoch": 0.3056768558951965, + "grad_norm": 0.588350309953633, + "learning_rate": 4.9999535300278666e-05, + "loss": 0.3763, + "step": 700 + }, + { + "epoch": 0.3078602620087336, + "grad_norm": 0.5721216258057977, + "learning_rate": 4.999906737638528e-05, + "loss": 0.3928, + "step": 705 + }, + { + "epoch": 0.31004366812227074, + "grad_norm": 1.0000585963765807, + "learning_rate": 4.9998438104028226e-05, + "loss": 0.3926, + "step": 710 + }, + { + "epoch": 0.31222707423580787, + "grad_norm": 2.2978661693892355, + "learning_rate": 4.999764748726891e-05, + "loss": 0.385, + "step": 715 + }, + { + "epoch": 0.314410480349345, + "grad_norm": 1.8144056170150589, + "learning_rate": 4.9996695531210135e-05, + "loss": 0.4175, + "step": 720 + }, + { + "epoch": 0.3165938864628821, + "grad_norm": 0.7618077273846784, + "learning_rate": 4.9995582241996e-05, + "loss": 0.4076, + "step": 725 + }, + { + "epoch": 0.31877729257641924, + "grad_norm": 0.8570756785316501, + "learning_rate": 4.9994307626811865e-05, + "loss": 0.3912, + "step": 730 + }, + { + "epoch": 0.32096069868995636, + "grad_norm": 0.6575118539772534, + "learning_rate": 4.999287169388433e-05, + "loss": 0.3913, + "step": 735 + }, + { + "epoch": 0.3231441048034934, + "grad_norm": 0.7341799315850096, + "learning_rate": 4.9991274452481164e-05, + "loss": 0.4078, + "step": 740 + }, + { + "epoch": 0.32532751091703055, + "grad_norm": 0.6776721860026673, + "learning_rate": 4.9989515912911234e-05, + "loss": 0.3957, + "step": 745 + }, + { + "epoch": 0.32751091703056767, + "grad_norm": 0.6298219471990844, + "learning_rate": 4.998759608652449e-05, + "loss": 0.4049, + "step": 750 + }, + { + "epoch": 0.3296943231441048, + "grad_norm": 0.5126548037982753, + "learning_rate": 4.998551498571182e-05, + "loss": 0.3929, + "step": 755 + }, + { + "epoch": 0.3318777292576419, + "grad_norm": 0.6814430013566892, + "learning_rate": 4.9983272623905045e-05, + "loss": 0.3823, + "step": 760 + }, + { + "epoch": 0.33406113537117904, + "grad_norm": 0.5643156159588145, + "learning_rate": 4.9980869015576736e-05, + "loss": 0.3761, + "step": 765 + }, + { + "epoch": 0.33624454148471616, + "grad_norm": 0.4499045190804366, + "learning_rate": 4.997830417624024e-05, + "loss": 0.3867, + "step": 770 + }, + { + "epoch": 0.3384279475982533, + "grad_norm": 0.7955611854097291, + "learning_rate": 4.9975578122449485e-05, + "loss": 0.3953, + "step": 775 + }, + { + "epoch": 0.3406113537117904, + "grad_norm": 0.5150140528502942, + "learning_rate": 4.99726908717989e-05, + "loss": 0.3769, + "step": 780 + }, + { + "epoch": 0.34279475982532753, + "grad_norm": 0.5896604319461464, + "learning_rate": 4.9969642442923325e-05, + "loss": 0.3776, + "step": 785 + }, + { + "epoch": 0.34497816593886466, + "grad_norm": 0.7333319728063213, + "learning_rate": 4.996643285549787e-05, + "loss": 0.3912, + "step": 790 + }, + { + "epoch": 0.3471615720524017, + "grad_norm": 0.975012641228752, + "learning_rate": 4.996306213023778e-05, + "loss": 0.374, + "step": 795 + }, + { + "epoch": 0.34934497816593885, + "grad_norm": 0.6363914305855141, + "learning_rate": 4.995953028889833e-05, + "loss": 0.3908, + "step": 800 + }, + { + "epoch": 0.35152838427947597, + "grad_norm": 0.7256711049149246, + "learning_rate": 4.995583735427465e-05, + "loss": 0.3804, + "step": 805 + }, + { + "epoch": 0.3537117903930131, + "grad_norm": 0.5201509210204703, + "learning_rate": 4.9951983350201604e-05, + "loss": 0.3787, + "step": 810 + }, + { + "epoch": 0.3558951965065502, + "grad_norm": 0.5937818267570265, + "learning_rate": 4.9947968301553625e-05, + "loss": 0.3681, + "step": 815 + }, + { + "epoch": 0.35807860262008734, + "grad_norm": 0.5805445788983403, + "learning_rate": 4.994379223424456e-05, + "loss": 0.3632, + "step": 820 + }, + { + "epoch": 0.36026200873362446, + "grad_norm": 0.4804762583689749, + "learning_rate": 4.99394551752275e-05, + "loss": 0.3729, + "step": 825 + }, + { + "epoch": 0.3624454148471616, + "grad_norm": 0.4780935144472406, + "learning_rate": 4.9934957152494585e-05, + "loss": 0.366, + "step": 830 + }, + { + "epoch": 0.3646288209606987, + "grad_norm": 0.46427385625879564, + "learning_rate": 4.993029819507687e-05, + "loss": 0.3641, + "step": 835 + }, + { + "epoch": 0.36681222707423583, + "grad_norm": 0.5283252072301436, + "learning_rate": 4.9925478333044114e-05, + "loss": 0.3797, + "step": 840 + }, + { + "epoch": 0.36899563318777295, + "grad_norm": 0.5593669993363974, + "learning_rate": 4.992049759750455e-05, + "loss": 0.3776, + "step": 845 + }, + { + "epoch": 0.37117903930131, + "grad_norm": 0.6282087870379117, + "learning_rate": 4.9915356020604754e-05, + "loss": 0.3723, + "step": 850 + }, + { + "epoch": 0.37336244541484714, + "grad_norm": 0.4876112040234335, + "learning_rate": 4.9910053635529376e-05, + "loss": 0.3918, + "step": 855 + }, + { + "epoch": 0.37554585152838427, + "grad_norm": 0.4642413234429665, + "learning_rate": 4.9904590476500955e-05, + "loss": 0.3809, + "step": 860 + }, + { + "epoch": 0.3777292576419214, + "grad_norm": 0.44822878442026826, + "learning_rate": 4.989896657877972e-05, + "loss": 0.37, + "step": 865 + }, + { + "epoch": 0.3799126637554585, + "grad_norm": 0.4846478506665109, + "learning_rate": 4.98931819786633e-05, + "loss": 0.3742, + "step": 870 + }, + { + "epoch": 0.38209606986899564, + "grad_norm": 0.4166354564636158, + "learning_rate": 4.988723671348654e-05, + "loss": 0.3658, + "step": 875 + }, + { + "epoch": 0.38427947598253276, + "grad_norm": 0.49223225972882195, + "learning_rate": 4.9881130821621255e-05, + "loss": 0.3914, + "step": 880 + }, + { + "epoch": 0.3864628820960699, + "grad_norm": 0.526884770677872, + "learning_rate": 4.987486434247596e-05, + "loss": 0.3754, + "step": 885 + }, + { + "epoch": 0.388646288209607, + "grad_norm": 0.5232997418172833, + "learning_rate": 4.9868437316495636e-05, + "loss": 0.3724, + "step": 890 + }, + { + "epoch": 0.39082969432314413, + "grad_norm": 0.48376676230007215, + "learning_rate": 4.9861849785161465e-05, + "loss": 0.3673, + "step": 895 + }, + { + "epoch": 0.3930131004366812, + "grad_norm": 0.5294590310273921, + "learning_rate": 4.9855101790990556e-05, + "loss": 0.3727, + "step": 900 + }, + { + "epoch": 0.3951965065502183, + "grad_norm": 0.47896380765481217, + "learning_rate": 4.9848193377535665e-05, + "loss": 0.3841, + "step": 905 + }, + { + "epoch": 0.39737991266375544, + "grad_norm": 0.607838261659882, + "learning_rate": 4.9841124589384944e-05, + "loss": 0.3723, + "step": 910 + }, + { + "epoch": 0.39956331877729256, + "grad_norm": 0.5096948643379262, + "learning_rate": 4.983389547216161e-05, + "loss": 0.3707, + "step": 915 + }, + { + "epoch": 0.4017467248908297, + "grad_norm": 0.38304781823406214, + "learning_rate": 4.9826506072523684e-05, + "loss": 0.3545, + "step": 920 + }, + { + "epoch": 0.4039301310043668, + "grad_norm": 0.587506061988026, + "learning_rate": 4.981895643816367e-05, + "loss": 0.3718, + "step": 925 + }, + { + "epoch": 0.40611353711790393, + "grad_norm": 0.5967719221364212, + "learning_rate": 4.9811246617808275e-05, + "loss": 0.3684, + "step": 930 + }, + { + "epoch": 0.40829694323144106, + "grad_norm": 0.6935599142313973, + "learning_rate": 4.980337666121806e-05, + "loss": 0.3789, + "step": 935 + }, + { + "epoch": 0.4104803493449782, + "grad_norm": 0.5766117112149044, + "learning_rate": 4.979534661918712e-05, + "loss": 0.3542, + "step": 940 + }, + { + "epoch": 0.4126637554585153, + "grad_norm": 0.43795584226862594, + "learning_rate": 4.978715654354281e-05, + "loss": 0.3595, + "step": 945 + }, + { + "epoch": 0.4148471615720524, + "grad_norm": 0.4430509318673314, + "learning_rate": 4.9778806487145326e-05, + "loss": 0.3666, + "step": 950 + }, + { + "epoch": 0.4170305676855895, + "grad_norm": 0.4481773430888534, + "learning_rate": 4.977029650388744e-05, + "loss": 0.3654, + "step": 955 + }, + { + "epoch": 0.4192139737991266, + "grad_norm": 0.4657407424462558, + "learning_rate": 4.976162664869411e-05, + "loss": 0.3542, + "step": 960 + }, + { + "epoch": 0.42139737991266374, + "grad_norm": 0.5287646283190662, + "learning_rate": 4.975279697752213e-05, + "loss": 0.3594, + "step": 965 + }, + { + "epoch": 0.42358078602620086, + "grad_norm": 0.4887455966923865, + "learning_rate": 4.974380754735979e-05, + "loss": 0.3669, + "step": 970 + }, + { + "epoch": 0.425764192139738, + "grad_norm": 0.43872446154490324, + "learning_rate": 4.9734658416226476e-05, + "loss": 0.3616, + "step": 975 + }, + { + "epoch": 0.4279475982532751, + "grad_norm": 0.6424735520245978, + "learning_rate": 4.972534964317233e-05, + "loss": 0.369, + "step": 980 + }, + { + "epoch": 0.43013100436681223, + "grad_norm": 0.48958723865804354, + "learning_rate": 4.9715881288277835e-05, + "loss": 0.3594, + "step": 985 + }, + { + "epoch": 0.43231441048034935, + "grad_norm": 0.4479761879052664, + "learning_rate": 4.9706253412653446e-05, + "loss": 0.3549, + "step": 990 + }, + { + "epoch": 0.4344978165938865, + "grad_norm": 0.5435346773895244, + "learning_rate": 4.9696466078439216e-05, + "loss": 0.3671, + "step": 995 + }, + { + "epoch": 0.4366812227074236, + "grad_norm": 0.5308581082448025, + "learning_rate": 4.968651934880434e-05, + "loss": 0.365, + "step": 1000 + }, + { + "epoch": 0.4388646288209607, + "grad_norm": 0.39896740398055186, + "learning_rate": 4.96764132879468e-05, + "loss": 0.3598, + "step": 1005 + }, + { + "epoch": 0.4410480349344978, + "grad_norm": 0.5067575654052805, + "learning_rate": 4.9666147961092924e-05, + "loss": 0.3659, + "step": 1010 + }, + { + "epoch": 0.4432314410480349, + "grad_norm": 0.5592601406097657, + "learning_rate": 4.9655723434496984e-05, + "loss": 0.3719, + "step": 1015 + }, + { + "epoch": 0.44541484716157204, + "grad_norm": 0.4413849291952618, + "learning_rate": 4.9645139775440716e-05, + "loss": 0.3652, + "step": 1020 + }, + { + "epoch": 0.44759825327510916, + "grad_norm": 0.3910244442044385, + "learning_rate": 4.9634397052232983e-05, + "loss": 0.3691, + "step": 1025 + }, + { + "epoch": 0.4497816593886463, + "grad_norm": 0.46071657374615793, + "learning_rate": 4.9623495334209226e-05, + "loss": 0.3624, + "step": 1030 + }, + { + "epoch": 0.4519650655021834, + "grad_norm": 0.4226092000790114, + "learning_rate": 4.96124346917311e-05, + "loss": 0.3676, + "step": 1035 + }, + { + "epoch": 0.45414847161572053, + "grad_norm": 0.44419459860035787, + "learning_rate": 4.960121519618597e-05, + "loss": 0.332, + "step": 1040 + }, + { + "epoch": 0.45633187772925765, + "grad_norm": 0.3577954443167031, + "learning_rate": 4.9589836919986475e-05, + "loss": 0.3512, + "step": 1045 + }, + { + "epoch": 0.4585152838427948, + "grad_norm": 0.5820450329129828, + "learning_rate": 4.957829993657005e-05, + "loss": 0.3501, + "step": 1050 + }, + { + "epoch": 0.4606986899563319, + "grad_norm": 0.5150191150181325, + "learning_rate": 4.956660432039846e-05, + "loss": 0.3597, + "step": 1055 + }, + { + "epoch": 0.462882096069869, + "grad_norm": 0.4319371225817441, + "learning_rate": 4.955475014695729e-05, + "loss": 0.3676, + "step": 1060 + }, + { + "epoch": 0.4650655021834061, + "grad_norm": 0.5533636569254564, + "learning_rate": 4.954273749275552e-05, + "loss": 0.3727, + "step": 1065 + }, + { + "epoch": 0.4672489082969432, + "grad_norm": 0.4958192610268456, + "learning_rate": 4.9530566435324963e-05, + "loss": 0.3488, + "step": 1070 + }, + { + "epoch": 0.46943231441048033, + "grad_norm": 0.5108202025863451, + "learning_rate": 4.9518237053219816e-05, + "loss": 0.3709, + "step": 1075 + }, + { + "epoch": 0.47161572052401746, + "grad_norm": 0.45372615060906546, + "learning_rate": 4.950574942601611e-05, + "loss": 0.3475, + "step": 1080 + }, + { + "epoch": 0.4737991266375546, + "grad_norm": 0.43181104589299185, + "learning_rate": 4.9493103634311245e-05, + "loss": 0.3527, + "step": 1085 + }, + { + "epoch": 0.4759825327510917, + "grad_norm": 0.49622919369152685, + "learning_rate": 4.948029975972341e-05, + "loss": 0.34, + "step": 1090 + }, + { + "epoch": 0.4781659388646288, + "grad_norm": 0.5352955894735107, + "learning_rate": 4.9467337884891127e-05, + "loss": 0.3578, + "step": 1095 + }, + { + "epoch": 0.48034934497816595, + "grad_norm": 0.48767956685977704, + "learning_rate": 4.945421809347266e-05, + "loss": 0.3437, + "step": 1100 + }, + { + "epoch": 0.48253275109170307, + "grad_norm": 0.5172947775347728, + "learning_rate": 4.944094047014547e-05, + "loss": 0.3582, + "step": 1105 + }, + { + "epoch": 0.4847161572052402, + "grad_norm": 0.4633995942526625, + "learning_rate": 4.942750510060574e-05, + "loss": 0.3369, + "step": 1110 + }, + { + "epoch": 0.4868995633187773, + "grad_norm": 0.5384683843923247, + "learning_rate": 4.9413912071567756e-05, + "loss": 0.3683, + "step": 1115 + }, + { + "epoch": 0.4890829694323144, + "grad_norm": 0.428948742370993, + "learning_rate": 4.940016147076336e-05, + "loss": 0.3425, + "step": 1120 + }, + { + "epoch": 0.4912663755458515, + "grad_norm": 0.5160388037360805, + "learning_rate": 4.938625338694139e-05, + "loss": 0.3609, + "step": 1125 + }, + { + "epoch": 0.49344978165938863, + "grad_norm": 0.45537160535635335, + "learning_rate": 4.937218790986712e-05, + "loss": 0.3532, + "step": 1130 + }, + { + "epoch": 0.49563318777292575, + "grad_norm": 0.4996736665030406, + "learning_rate": 4.935796513032165e-05, + "loss": 0.3434, + "step": 1135 + }, + { + "epoch": 0.4978165938864629, + "grad_norm": 0.5155894229667807, + "learning_rate": 4.934358514010137e-05, + "loss": 0.3569, + "step": 1140 + }, + { + "epoch": 0.5, + "grad_norm": 0.38793250116768685, + "learning_rate": 4.93290480320173e-05, + "loss": 0.3545, + "step": 1145 + }, + { + "epoch": 0.5021834061135371, + "grad_norm": 0.36522790796535515, + "learning_rate": 4.931435389989455e-05, + "loss": 0.3482, + "step": 1150 + }, + { + "epoch": 0.5043668122270742, + "grad_norm": 0.46762083247492375, + "learning_rate": 4.929950283857167e-05, + "loss": 0.3704, + "step": 1155 + }, + { + "epoch": 0.5065502183406113, + "grad_norm": 0.45795606465334227, + "learning_rate": 4.9284494943900085e-05, + "loss": 0.3646, + "step": 1160 + }, + { + "epoch": 0.5087336244541485, + "grad_norm": 0.42995508768886065, + "learning_rate": 4.926933031274344e-05, + "loss": 0.3626, + "step": 1165 + }, + { + "epoch": 0.5109170305676856, + "grad_norm": 0.4224390944477537, + "learning_rate": 4.925400904297698e-05, + "loss": 0.347, + "step": 1170 + }, + { + "epoch": 0.5131004366812227, + "grad_norm": 0.5046566006479356, + "learning_rate": 4.923853123348693e-05, + "loss": 0.3678, + "step": 1175 + }, + { + "epoch": 0.5152838427947598, + "grad_norm": 0.5310348130320324, + "learning_rate": 4.922289698416983e-05, + "loss": 0.3568, + "step": 1180 + }, + { + "epoch": 0.517467248908297, + "grad_norm": 0.44019956240689734, + "learning_rate": 4.920710639593196e-05, + "loss": 0.3555, + "step": 1185 + }, + { + "epoch": 0.519650655021834, + "grad_norm": 0.36638621076470246, + "learning_rate": 4.919115957068859e-05, + "loss": 0.3529, + "step": 1190 + }, + { + "epoch": 0.5218340611353712, + "grad_norm": 0.4899606255024857, + "learning_rate": 4.917505661136339e-05, + "loss": 0.3491, + "step": 1195 + }, + { + "epoch": 0.5240174672489083, + "grad_norm": 0.501601743259068, + "learning_rate": 4.915879762188774e-05, + "loss": 0.3615, + "step": 1200 + }, + { + "epoch": 0.5262008733624454, + "grad_norm": 0.6025124136368822, + "learning_rate": 4.9142382707200096e-05, + "loss": 0.3553, + "step": 1205 + }, + { + "epoch": 0.5283842794759825, + "grad_norm": 0.445329859368048, + "learning_rate": 4.912581197324524e-05, + "loss": 0.3633, + "step": 1210 + }, + { + "epoch": 0.5305676855895196, + "grad_norm": 0.6418024421311272, + "learning_rate": 4.9109085526973674e-05, + "loss": 0.3476, + "step": 1215 + }, + { + "epoch": 0.5327510917030568, + "grad_norm": 0.4430605711065509, + "learning_rate": 4.909220347634087e-05, + "loss": 0.359, + "step": 1220 + }, + { + "epoch": 0.5349344978165939, + "grad_norm": 0.4354421223792483, + "learning_rate": 4.907516593030662e-05, + "loss": 0.3515, + "step": 1225 + }, + { + "epoch": 0.537117903930131, + "grad_norm": 0.47281124382089545, + "learning_rate": 4.905797299883429e-05, + "loss": 0.3629, + "step": 1230 + }, + { + "epoch": 0.5393013100436681, + "grad_norm": 0.4468031895316623, + "learning_rate": 4.904062479289015e-05, + "loss": 0.3484, + "step": 1235 + }, + { + "epoch": 0.5414847161572053, + "grad_norm": 0.4659670701752814, + "learning_rate": 4.902312142444264e-05, + "loss": 0.3526, + "step": 1240 + }, + { + "epoch": 0.5436681222707423, + "grad_norm": 0.371869503352742, + "learning_rate": 4.900546300646162e-05, + "loss": 0.3366, + "step": 1245 + }, + { + "epoch": 0.5458515283842795, + "grad_norm": 0.42423288318858376, + "learning_rate": 4.89876496529177e-05, + "loss": 0.3543, + "step": 1250 + }, + { + "epoch": 0.5480349344978166, + "grad_norm": 0.4609253407888565, + "learning_rate": 4.896968147878145e-05, + "loss": 0.3565, + "step": 1255 + }, + { + "epoch": 0.5502183406113537, + "grad_norm": 0.3237267708873633, + "learning_rate": 4.895155860002271e-05, + "loss": 0.3377, + "step": 1260 + }, + { + "epoch": 0.5524017467248908, + "grad_norm": 0.4546639718401833, + "learning_rate": 4.893328113360975e-05, + "loss": 0.3535, + "step": 1265 + }, + { + "epoch": 0.5545851528384279, + "grad_norm": 0.43148670011250856, + "learning_rate": 4.891484919750865e-05, + "loss": 0.344, + "step": 1270 + }, + { + "epoch": 0.5567685589519651, + "grad_norm": 0.3991105241595635, + "learning_rate": 4.889626291068241e-05, + "loss": 0.339, + "step": 1275 + }, + { + "epoch": 0.5589519650655022, + "grad_norm": 0.48987054857422774, + "learning_rate": 4.887752239309027e-05, + "loss": 0.3391, + "step": 1280 + }, + { + "epoch": 0.5611353711790393, + "grad_norm": 0.4609741215989259, + "learning_rate": 4.885862776568689e-05, + "loss": 0.3576, + "step": 1285 + }, + { + "epoch": 0.5633187772925764, + "grad_norm": 0.50704867138633, + "learning_rate": 4.883957915042158e-05, + "loss": 0.3496, + "step": 1290 + }, + { + "epoch": 0.5655021834061136, + "grad_norm": 0.48159077762660746, + "learning_rate": 4.882037667023752e-05, + "loss": 0.3361, + "step": 1295 + }, + { + "epoch": 0.5676855895196506, + "grad_norm": 0.7062981902269905, + "learning_rate": 4.880102044907096e-05, + "loss": 0.3677, + "step": 1300 + }, + { + "epoch": 0.5698689956331878, + "grad_norm": 0.4888521580220186, + "learning_rate": 4.878151061185042e-05, + "loss": 0.3294, + "step": 1305 + }, + { + "epoch": 0.5720524017467249, + "grad_norm": 0.4540470014267252, + "learning_rate": 4.876184728449589e-05, + "loss": 0.3488, + "step": 1310 + }, + { + "epoch": 0.574235807860262, + "grad_norm": 0.3930373040922702, + "learning_rate": 4.874203059391802e-05, + "loss": 0.3364, + "step": 1315 + }, + { + "epoch": 0.5764192139737991, + "grad_norm": 0.4451157942220829, + "learning_rate": 4.8722060668017267e-05, + "loss": 0.3562, + "step": 1320 + }, + { + "epoch": 0.5786026200873362, + "grad_norm": 0.3942536599077881, + "learning_rate": 4.870193763568312e-05, + "loss": 0.3619, + "step": 1325 + }, + { + "epoch": 0.5807860262008734, + "grad_norm": 0.48861179907906854, + "learning_rate": 4.8681661626793246e-05, + "loss": 0.3468, + "step": 1330 + }, + { + "epoch": 0.5829694323144105, + "grad_norm": 0.42138612451614066, + "learning_rate": 4.866123277221264e-05, + "loss": 0.347, + "step": 1335 + }, + { + "epoch": 0.5851528384279476, + "grad_norm": 0.4074290406551753, + "learning_rate": 4.864065120379278e-05, + "loss": 0.3573, + "step": 1340 + }, + { + "epoch": 0.5873362445414847, + "grad_norm": 0.3987198118209572, + "learning_rate": 4.861991705437081e-05, + "loss": 0.3598, + "step": 1345 + }, + { + "epoch": 0.5895196506550219, + "grad_norm": 0.46460108590367194, + "learning_rate": 4.859903045776866e-05, + "loss": 0.3524, + "step": 1350 + }, + { + "epoch": 0.5917030567685589, + "grad_norm": 0.4803498793618286, + "learning_rate": 4.857799154879215e-05, + "loss": 0.3605, + "step": 1355 + }, + { + "epoch": 0.5938864628820961, + "grad_norm": 0.5901063012029109, + "learning_rate": 4.8556800463230165e-05, + "loss": 0.3437, + "step": 1360 + }, + { + "epoch": 0.5960698689956332, + "grad_norm": 0.4641267585126181, + "learning_rate": 4.8535457337853794e-05, + "loss": 0.3285, + "step": 1365 + }, + { + "epoch": 0.5982532751091703, + "grad_norm": 0.3647796263943895, + "learning_rate": 4.8513962310415396e-05, + "loss": 0.3463, + "step": 1370 + }, + { + "epoch": 0.6004366812227074, + "grad_norm": 0.3648828726715139, + "learning_rate": 4.849231551964771e-05, + "loss": 0.3535, + "step": 1375 + }, + { + "epoch": 0.6026200873362445, + "grad_norm": 0.40542343011346316, + "learning_rate": 4.847051710526302e-05, + "loss": 0.3437, + "step": 1380 + }, + { + "epoch": 0.6048034934497817, + "grad_norm": 0.350582105718404, + "learning_rate": 4.8448567207952205e-05, + "loss": 0.3549, + "step": 1385 + }, + { + "epoch": 0.6069868995633187, + "grad_norm": 0.39947990507599496, + "learning_rate": 4.8426465969383826e-05, + "loss": 0.3243, + "step": 1390 + }, + { + "epoch": 0.6091703056768559, + "grad_norm": 0.4129694646452404, + "learning_rate": 4.840421353220326e-05, + "loss": 0.3251, + "step": 1395 + }, + { + "epoch": 0.611353711790393, + "grad_norm": 0.3820909710575323, + "learning_rate": 4.83818100400317e-05, + "loss": 0.334, + "step": 1400 + }, + { + "epoch": 0.6135371179039302, + "grad_norm": 0.4069293886007247, + "learning_rate": 4.835925563746532e-05, + "loss": 0.3551, + "step": 1405 + }, + { + "epoch": 0.6157205240174672, + "grad_norm": 0.4582430996524103, + "learning_rate": 4.833655047007427e-05, + "loss": 0.3504, + "step": 1410 + }, + { + "epoch": 0.6179039301310044, + "grad_norm": 0.3828069417511488, + "learning_rate": 4.831369468440177e-05, + "loss": 0.3358, + "step": 1415 + }, + { + "epoch": 0.6200873362445415, + "grad_norm": 0.379141828052868, + "learning_rate": 4.8290688427963175e-05, + "loss": 0.3315, + "step": 1420 + }, + { + "epoch": 0.6222707423580786, + "grad_norm": 0.4059377645684439, + "learning_rate": 4.8267531849244974e-05, + "loss": 0.3403, + "step": 1425 + }, + { + "epoch": 0.6244541484716157, + "grad_norm": 0.3781489511039065, + "learning_rate": 4.8244225097703885e-05, + "loss": 0.351, + "step": 1430 + }, + { + "epoch": 0.6266375545851528, + "grad_norm": 0.36811899180808383, + "learning_rate": 4.8220768323765866e-05, + "loss": 0.345, + "step": 1435 + }, + { + "epoch": 0.62882096069869, + "grad_norm": 0.48695298818921573, + "learning_rate": 4.819716167882513e-05, + "loss": 0.343, + "step": 1440 + }, + { + "epoch": 0.631004366812227, + "grad_norm": 0.41424246411412347, + "learning_rate": 4.8173405315243216e-05, + "loss": 0.3381, + "step": 1445 + }, + { + "epoch": 0.6331877729257642, + "grad_norm": 0.5982905072375967, + "learning_rate": 4.814949938634793e-05, + "loss": 0.3334, + "step": 1450 + }, + { + "epoch": 0.6353711790393013, + "grad_norm": 0.41432212938369534, + "learning_rate": 4.8125444046432434e-05, + "loss": 0.3387, + "step": 1455 + }, + { + "epoch": 0.6375545851528385, + "grad_norm": 0.3970797596419897, + "learning_rate": 4.810123945075421e-05, + "loss": 0.3485, + "step": 1460 + }, + { + "epoch": 0.6397379912663755, + "grad_norm": 0.36649659825973896, + "learning_rate": 4.807688575553406e-05, + "loss": 0.3381, + "step": 1465 + }, + { + "epoch": 0.6419213973799127, + "grad_norm": 0.4151615051412959, + "learning_rate": 4.80523831179551e-05, + "loss": 0.3316, + "step": 1470 + }, + { + "epoch": 0.6441048034934498, + "grad_norm": 0.41296409849005844, + "learning_rate": 4.8027731696161735e-05, + "loss": 0.3378, + "step": 1475 + }, + { + "epoch": 0.6462882096069869, + "grad_norm": 0.43599613227274564, + "learning_rate": 4.8002931649258674e-05, + "loss": 0.3343, + "step": 1480 + }, + { + "epoch": 0.648471615720524, + "grad_norm": 0.5041455481985673, + "learning_rate": 4.7977983137309855e-05, + "loss": 0.3337, + "step": 1485 + }, + { + "epoch": 0.6506550218340611, + "grad_norm": 0.41603606102503005, + "learning_rate": 4.795288632133747e-05, + "loss": 0.3328, + "step": 1490 + }, + { + "epoch": 0.6528384279475983, + "grad_norm": 0.4091756291458897, + "learning_rate": 4.792764136332084e-05, + "loss": 0.3558, + "step": 1495 + }, + { + "epoch": 0.6550218340611353, + "grad_norm": 0.3535456736539121, + "learning_rate": 4.790224842619548e-05, + "loss": 0.3305, + "step": 1500 + }, + { + "epoch": 0.6572052401746725, + "grad_norm": 0.4204136805374324, + "learning_rate": 4.787670767385194e-05, + "loss": 0.3399, + "step": 1505 + }, + { + "epoch": 0.6593886462882096, + "grad_norm": 0.4138703097114033, + "learning_rate": 4.785101927113482e-05, + "loss": 0.3423, + "step": 1510 + }, + { + "epoch": 0.6615720524017468, + "grad_norm": 0.42867964938598574, + "learning_rate": 4.782518338384168e-05, + "loss": 0.3474, + "step": 1515 + }, + { + "epoch": 0.6637554585152838, + "grad_norm": 0.3655864787630958, + "learning_rate": 4.779920017872197e-05, + "loss": 0.3426, + "step": 1520 + }, + { + "epoch": 0.665938864628821, + "grad_norm": 0.31312668856989573, + "learning_rate": 4.7773069823475945e-05, + "loss": 0.3317, + "step": 1525 + }, + { + "epoch": 0.6681222707423581, + "grad_norm": 0.3720008834724695, + "learning_rate": 4.774679248675362e-05, + "loss": 0.3468, + "step": 1530 + }, + { + "epoch": 0.6703056768558951, + "grad_norm": 0.3500906347775602, + "learning_rate": 4.772036833815363e-05, + "loss": 0.3305, + "step": 1535 + }, + { + "epoch": 0.6724890829694323, + "grad_norm": 0.4324016765021572, + "learning_rate": 4.7693797548222166e-05, + "loss": 0.3226, + "step": 1540 + }, + { + "epoch": 0.6746724890829694, + "grad_norm": 0.40696438387223544, + "learning_rate": 4.766708028845188e-05, + "loss": 0.3325, + "step": 1545 + }, + { + "epoch": 0.6768558951965066, + "grad_norm": 0.4155361411655341, + "learning_rate": 4.764021673128075e-05, + "loss": 0.3482, + "step": 1550 + }, + { + "epoch": 0.6790393013100436, + "grad_norm": 0.4890226077260248, + "learning_rate": 4.761320705009099e-05, + "loss": 0.3457, + "step": 1555 + }, + { + "epoch": 0.6812227074235808, + "grad_norm": 0.42425030996602786, + "learning_rate": 4.758605141920792e-05, + "loss": 0.3512, + "step": 1560 + }, + { + "epoch": 0.6834061135371179, + "grad_norm": 0.3594877731084044, + "learning_rate": 4.755875001389887e-05, + "loss": 0.3434, + "step": 1565 + }, + { + "epoch": 0.6855895196506551, + "grad_norm": 0.3979846319153574, + "learning_rate": 4.753130301037199e-05, + "loss": 0.3337, + "step": 1570 + }, + { + "epoch": 0.6877729257641921, + "grad_norm": 0.402834791445396, + "learning_rate": 4.7503710585775174e-05, + "loss": 0.3387, + "step": 1575 + }, + { + "epoch": 0.6899563318777293, + "grad_norm": 0.34220820957249626, + "learning_rate": 4.747597291819487e-05, + "loss": 0.3439, + "step": 1580 + }, + { + "epoch": 0.6921397379912664, + "grad_norm": 0.3831695794480096, + "learning_rate": 4.744809018665495e-05, + "loss": 0.3373, + "step": 1585 + }, + { + "epoch": 0.6943231441048034, + "grad_norm": 0.40957271083541796, + "learning_rate": 4.742006257111558e-05, + "loss": 0.3358, + "step": 1590 + }, + { + "epoch": 0.6965065502183406, + "grad_norm": 0.3989291887534643, + "learning_rate": 4.7391890252472016e-05, + "loss": 0.3624, + "step": 1595 + }, + { + "epoch": 0.6986899563318777, + "grad_norm": 0.47006740071344755, + "learning_rate": 4.736357341255344e-05, + "loss": 0.3427, + "step": 1600 + }, + { + "epoch": 0.7008733624454149, + "grad_norm": 0.5495070509701956, + "learning_rate": 4.7335112234121824e-05, + "loss": 0.3332, + "step": 1605 + }, + { + "epoch": 0.7030567685589519, + "grad_norm": 0.37491891454833864, + "learning_rate": 4.730650690087072e-05, + "loss": 0.3406, + "step": 1610 + }, + { + "epoch": 0.7052401746724891, + "grad_norm": 0.43036507029194554, + "learning_rate": 4.727775759742408e-05, + "loss": 0.3431, + "step": 1615 + }, + { + "epoch": 0.7074235807860262, + "grad_norm": 0.3820782925001961, + "learning_rate": 4.724886450933505e-05, + "loss": 0.3436, + "step": 1620 + }, + { + "epoch": 0.7096069868995634, + "grad_norm": 0.4860255736060773, + "learning_rate": 4.721982782308481e-05, + "loss": 0.3523, + "step": 1625 + }, + { + "epoch": 0.7117903930131004, + "grad_norm": 0.3695141770304634, + "learning_rate": 4.719064772608134e-05, + "loss": 0.3462, + "step": 1630 + }, + { + "epoch": 0.7139737991266376, + "grad_norm": 0.4003960638271693, + "learning_rate": 4.7161324406658225e-05, + "loss": 0.3286, + "step": 1635 + }, + { + "epoch": 0.7161572052401747, + "grad_norm": 0.3530404563540438, + "learning_rate": 4.713185805407342e-05, + "loss": 0.3236, + "step": 1640 + }, + { + "epoch": 0.7183406113537117, + "grad_norm": 0.584075905055993, + "learning_rate": 4.7102248858508044e-05, + "loss": 0.3544, + "step": 1645 + }, + { + "epoch": 0.7205240174672489, + "grad_norm": 0.46166165713972934, + "learning_rate": 4.7072497011065156e-05, + "loss": 0.3298, + "step": 1650 + }, + { + "epoch": 0.722707423580786, + "grad_norm": 0.3355056455618887, + "learning_rate": 4.704260270376853e-05, + "loss": 0.3264, + "step": 1655 + }, + { + "epoch": 0.7248908296943232, + "grad_norm": 0.39311481965484996, + "learning_rate": 4.7012566129561365e-05, + "loss": 0.3381, + "step": 1660 + }, + { + "epoch": 0.7270742358078602, + "grad_norm": 0.38148657969468514, + "learning_rate": 4.6982387482305104e-05, + "loss": 0.3322, + "step": 1665 + }, + { + "epoch": 0.7292576419213974, + "grad_norm": 0.44427131750278887, + "learning_rate": 4.6952066956778155e-05, + "loss": 0.3268, + "step": 1670 + }, + { + "epoch": 0.7314410480349345, + "grad_norm": 0.3857743094672692, + "learning_rate": 4.692160474867463e-05, + "loss": 0.3324, + "step": 1675 + }, + { + "epoch": 0.7336244541484717, + "grad_norm": 0.3899786180476667, + "learning_rate": 4.6891001054603056e-05, + "loss": 0.3374, + "step": 1680 + }, + { + "epoch": 0.7358078602620087, + "grad_norm": 0.3696166334234957, + "learning_rate": 4.6860256072085196e-05, + "loss": 0.3453, + "step": 1685 + }, + { + "epoch": 0.7379912663755459, + "grad_norm": 0.30167847925632657, + "learning_rate": 4.6829369999554664e-05, + "loss": 0.3297, + "step": 1690 + }, + { + "epoch": 0.740174672489083, + "grad_norm": 0.37202185004193333, + "learning_rate": 4.6798343036355704e-05, + "loss": 0.3058, + "step": 1695 + }, + { + "epoch": 0.74235807860262, + "grad_norm": 0.3965117699988383, + "learning_rate": 4.6767175382741914e-05, + "loss": 0.3367, + "step": 1700 + }, + { + "epoch": 0.7445414847161572, + "grad_norm": 0.37265664937861914, + "learning_rate": 4.673586723987491e-05, + "loss": 0.3504, + "step": 1705 + }, + { + "epoch": 0.7467248908296943, + "grad_norm": 0.42824465747750634, + "learning_rate": 4.670441880982307e-05, + "loss": 0.3323, + "step": 1710 + }, + { + "epoch": 0.7489082969432315, + "grad_norm": 0.3708258588256483, + "learning_rate": 4.667283029556019e-05, + "loss": 0.3288, + "step": 1715 + }, + { + "epoch": 0.7510917030567685, + "grad_norm": 0.4271790614273471, + "learning_rate": 4.664110190096421e-05, + "loss": 0.3436, + "step": 1720 + }, + { + "epoch": 0.7532751091703057, + "grad_norm": 0.37662914132162023, + "learning_rate": 4.660923383081588e-05, + "loss": 0.3401, + "step": 1725 + }, + { + "epoch": 0.7554585152838428, + "grad_norm": 0.3574814773268414, + "learning_rate": 4.657722629079744e-05, + "loss": 0.3359, + "step": 1730 + }, + { + "epoch": 0.75764192139738, + "grad_norm": 0.44811194231414175, + "learning_rate": 4.6545079487491285e-05, + "loss": 0.3466, + "step": 1735 + }, + { + "epoch": 0.759825327510917, + "grad_norm": 0.5433785036597635, + "learning_rate": 4.6512793628378657e-05, + "loss": 0.3348, + "step": 1740 + }, + { + "epoch": 0.7620087336244541, + "grad_norm": 0.424131707206492, + "learning_rate": 4.648036892183828e-05, + "loss": 0.3381, + "step": 1745 + }, + { + "epoch": 0.7641921397379913, + "grad_norm": 0.3485719470067596, + "learning_rate": 4.644780557714502e-05, + "loss": 0.3338, + "step": 1750 + }, + { + "epoch": 0.7663755458515283, + "grad_norm": 0.3595654663623888, + "learning_rate": 4.641510380446854e-05, + "loss": 0.3238, + "step": 1755 + }, + { + "epoch": 0.7685589519650655, + "grad_norm": 0.4262032500390272, + "learning_rate": 4.638226381487195e-05, + "loss": 0.3373, + "step": 1760 + }, + { + "epoch": 0.7707423580786026, + "grad_norm": 0.44078101888220983, + "learning_rate": 4.634928582031043e-05, + "loss": 0.3449, + "step": 1765 + }, + { + "epoch": 0.7729257641921398, + "grad_norm": 0.4054216447358104, + "learning_rate": 4.631617003362988e-05, + "loss": 0.3337, + "step": 1770 + }, + { + "epoch": 0.7751091703056768, + "grad_norm": 0.3848597633812537, + "learning_rate": 4.628291666856551e-05, + "loss": 0.3211, + "step": 1775 + }, + { + "epoch": 0.777292576419214, + "grad_norm": 0.3408957154658452, + "learning_rate": 4.6249525939740514e-05, + "loss": 0.3306, + "step": 1780 + }, + { + "epoch": 0.7794759825327511, + "grad_norm": 0.4050082743023021, + "learning_rate": 4.6215998062664636e-05, + "loss": 0.336, + "step": 1785 + }, + { + "epoch": 0.7816593886462883, + "grad_norm": 0.4303256630650177, + "learning_rate": 4.61823332537328e-05, + "loss": 0.3271, + "step": 1790 + }, + { + "epoch": 0.7838427947598253, + "grad_norm": 0.3879573361229342, + "learning_rate": 4.6148531730223735e-05, + "loss": 0.3309, + "step": 1795 + }, + { + "epoch": 0.7860262008733624, + "grad_norm": 0.3340993854538669, + "learning_rate": 4.611459371029853e-05, + "loss": 0.3273, + "step": 1800 + }, + { + "epoch": 0.7882096069868996, + "grad_norm": 0.41220645591308713, + "learning_rate": 4.6080519412999244e-05, + "loss": 0.3256, + "step": 1805 + }, + { + "epoch": 0.7903930131004366, + "grad_norm": 0.4607924969239619, + "learning_rate": 4.604630905824751e-05, + "loss": 0.3344, + "step": 1810 + }, + { + "epoch": 0.7925764192139738, + "grad_norm": 0.3986158317545471, + "learning_rate": 4.601196286684309e-05, + "loss": 0.3301, + "step": 1815 + }, + { + "epoch": 0.7947598253275109, + "grad_norm": 0.31961695415950014, + "learning_rate": 4.597748106046247e-05, + "loss": 0.3305, + "step": 1820 + }, + { + "epoch": 0.7969432314410481, + "grad_norm": 0.390531096769233, + "learning_rate": 4.5942863861657394e-05, + "loss": 0.3359, + "step": 1825 + }, + { + "epoch": 0.7991266375545851, + "grad_norm": 0.3431703506154723, + "learning_rate": 4.590811149385348e-05, + "loss": 0.3422, + "step": 1830 + }, + { + "epoch": 0.8013100436681223, + "grad_norm": 0.41192305785453975, + "learning_rate": 4.5873224181348736e-05, + "loss": 0.3247, + "step": 1835 + }, + { + "epoch": 0.8034934497816594, + "grad_norm": 0.2924629037593409, + "learning_rate": 4.5838202149312145e-05, + "loss": 0.3304, + "step": 1840 + }, + { + "epoch": 0.8056768558951966, + "grad_norm": 0.5634336476609664, + "learning_rate": 4.580304562378218e-05, + "loss": 0.3303, + "step": 1845 + }, + { + "epoch": 0.8078602620087336, + "grad_norm": 0.3544990534025258, + "learning_rate": 4.576775483166536e-05, + "loss": 0.3234, + "step": 1850 + }, + { + "epoch": 0.8100436681222707, + "grad_norm": 0.3459858077037921, + "learning_rate": 4.5732330000734775e-05, + "loss": 0.3319, + "step": 1855 + }, + { + "epoch": 0.8122270742358079, + "grad_norm": 0.3524884912271132, + "learning_rate": 4.569677135962864e-05, + "loss": 0.3278, + "step": 1860 + }, + { + "epoch": 0.8144104803493449, + "grad_norm": 0.42168404868370957, + "learning_rate": 4.5661079137848817e-05, + "loss": 0.314, + "step": 1865 + }, + { + "epoch": 0.8165938864628821, + "grad_norm": 0.4177554810132261, + "learning_rate": 4.562525356575928e-05, + "loss": 0.319, + "step": 1870 + }, + { + "epoch": 0.8187772925764192, + "grad_norm": 0.3406360499086555, + "learning_rate": 4.558929487458472e-05, + "loss": 0.3307, + "step": 1875 + }, + { + "epoch": 0.8209606986899564, + "grad_norm": 0.5953338048353011, + "learning_rate": 4.5553203296408966e-05, + "loss": 0.3363, + "step": 1880 + }, + { + "epoch": 0.8231441048034934, + "grad_norm": 0.3755495334670722, + "learning_rate": 4.551697906417353e-05, + "loss": 0.3228, + "step": 1885 + }, + { + "epoch": 0.8253275109170306, + "grad_norm": 0.5244192452428544, + "learning_rate": 4.5480622411676105e-05, + "loss": 0.3371, + "step": 1890 + }, + { + "epoch": 0.8275109170305677, + "grad_norm": 0.42718167146451547, + "learning_rate": 4.5444133573569056e-05, + "loss": 0.3394, + "step": 1895 + }, + { + "epoch": 0.8296943231441049, + "grad_norm": 0.5786026070553736, + "learning_rate": 4.540751278535788e-05, + "loss": 0.3413, + "step": 1900 + }, + { + "epoch": 0.8318777292576419, + "grad_norm": 0.3994255104166588, + "learning_rate": 4.5370760283399714e-05, + "loss": 0.3237, + "step": 1905 + }, + { + "epoch": 0.834061135371179, + "grad_norm": 0.465218260660747, + "learning_rate": 4.533387630490179e-05, + "loss": 0.3258, + "step": 1910 + }, + { + "epoch": 0.8362445414847162, + "grad_norm": 0.3503020390555209, + "learning_rate": 4.529686108791993e-05, + "loss": 0.3365, + "step": 1915 + }, + { + "epoch": 0.8384279475982532, + "grad_norm": 0.2937649633191906, + "learning_rate": 4.525971487135697e-05, + "loss": 0.3261, + "step": 1920 + }, + { + "epoch": 0.8406113537117904, + "grad_norm": 0.3640608075443947, + "learning_rate": 4.522243789496125e-05, + "loss": 0.3314, + "step": 1925 + }, + { + "epoch": 0.8427947598253275, + "grad_norm": 0.46642738830210834, + "learning_rate": 4.518503039932508e-05, + "loss": 0.3294, + "step": 1930 + }, + { + "epoch": 0.8449781659388647, + "grad_norm": 0.49375163248568255, + "learning_rate": 4.5147492625883133e-05, + "loss": 0.3346, + "step": 1935 + }, + { + "epoch": 0.8471615720524017, + "grad_norm": 0.43745545950003145, + "learning_rate": 4.510982481691093e-05, + "loss": 0.3356, + "step": 1940 + }, + { + "epoch": 0.8493449781659389, + "grad_norm": 0.3849745896242002, + "learning_rate": 4.5072027215523256e-05, + "loss": 0.328, + "step": 1945 + }, + { + "epoch": 0.851528384279476, + "grad_norm": 0.323723856108267, + "learning_rate": 4.503410006567262e-05, + "loss": 0.3137, + "step": 1950 + }, + { + "epoch": 0.8537117903930131, + "grad_norm": 0.3497043951204244, + "learning_rate": 4.499604361214764e-05, + "loss": 0.3367, + "step": 1955 + }, + { + "epoch": 0.8558951965065502, + "grad_norm": 0.3279206934648897, + "learning_rate": 4.495785810057151e-05, + "loss": 0.3305, + "step": 1960 + }, + { + "epoch": 0.8580786026200873, + "grad_norm": 0.35534435710755663, + "learning_rate": 4.4919543777400354e-05, + "loss": 0.3148, + "step": 1965 + }, + { + "epoch": 0.8602620087336245, + "grad_norm": 0.35737396766118557, + "learning_rate": 4.488110088992168e-05, + "loss": 0.3365, + "step": 1970 + }, + { + "epoch": 0.8624454148471615, + "grad_norm": 0.3811344665888169, + "learning_rate": 4.4842529686252774e-05, + "loss": 0.3412, + "step": 1975 + }, + { + "epoch": 0.8646288209606987, + "grad_norm": 0.3586035760646476, + "learning_rate": 4.4803830415339124e-05, + "loss": 0.3319, + "step": 1980 + }, + { + "epoch": 0.8668122270742358, + "grad_norm": 0.35951201381552045, + "learning_rate": 4.476500332695275e-05, + "loss": 0.3123, + "step": 1985 + }, + { + "epoch": 0.868995633187773, + "grad_norm": 0.4398889292154198, + "learning_rate": 4.472604867169063e-05, + "loss": 0.3258, + "step": 1990 + }, + { + "epoch": 0.87117903930131, + "grad_norm": 0.45684784821543106, + "learning_rate": 4.468696670097312e-05, + "loss": 0.3321, + "step": 1995 + }, + { + "epoch": 0.8733624454148472, + "grad_norm": 0.4624735089982277, + "learning_rate": 4.464775766704225e-05, + "loss": 0.339, + "step": 2000 + }, + { + "epoch": 0.8755458515283843, + "grad_norm": 0.39373793988338507, + "learning_rate": 4.4608421822960165e-05, + "loss": 0.3232, + "step": 2005 + }, + { + "epoch": 0.8777292576419214, + "grad_norm": 0.4793265990786189, + "learning_rate": 4.456895942260745e-05, + "loss": 0.3115, + "step": 2010 + }, + { + "epoch": 0.8799126637554585, + "grad_norm": 0.4596917524946721, + "learning_rate": 4.4529370720681516e-05, + "loss": 0.3327, + "step": 2015 + }, + { + "epoch": 0.8820960698689956, + "grad_norm": 0.41630714877910346, + "learning_rate": 4.4489655972694946e-05, + "loss": 0.3234, + "step": 2020 + }, + { + "epoch": 0.8842794759825328, + "grad_norm": 0.433659367959683, + "learning_rate": 4.444981543497384e-05, + "loss": 0.3268, + "step": 2025 + }, + { + "epoch": 0.8864628820960698, + "grad_norm": 0.4911763094493765, + "learning_rate": 4.440984936465618e-05, + "loss": 0.3283, + "step": 2030 + }, + { + "epoch": 0.888646288209607, + "grad_norm": 0.43509708304067723, + "learning_rate": 4.436975801969013e-05, + "loss": 0.3166, + "step": 2035 + }, + { + "epoch": 0.8908296943231441, + "grad_norm": 0.32181145619100665, + "learning_rate": 4.4329541658832454e-05, + "loss": 0.3275, + "step": 2040 + }, + { + "epoch": 0.8930131004366813, + "grad_norm": 0.5173193806980154, + "learning_rate": 4.428920054164673e-05, + "loss": 0.3367, + "step": 2045 + }, + { + "epoch": 0.8951965065502183, + "grad_norm": 0.31983525834019055, + "learning_rate": 4.424873492850177e-05, + "loss": 0.3332, + "step": 2050 + }, + { + "epoch": 0.8973799126637555, + "grad_norm": 0.3070335963036919, + "learning_rate": 4.42081450805699e-05, + "loss": 0.3416, + "step": 2055 + }, + { + "epoch": 0.8995633187772926, + "grad_norm": 0.33240317525320323, + "learning_rate": 4.4167431259825274e-05, + "loss": 0.3214, + "step": 2060 + }, + { + "epoch": 0.9017467248908297, + "grad_norm": 0.6261250657003081, + "learning_rate": 4.41265937290422e-05, + "loss": 0.3403, + "step": 2065 + }, + { + "epoch": 0.9039301310043668, + "grad_norm": 0.6469352618052645, + "learning_rate": 4.408563275179341e-05, + "loss": 0.3161, + "step": 2070 + }, + { + "epoch": 0.9061135371179039, + "grad_norm": 0.5128181692159853, + "learning_rate": 4.404454859244841e-05, + "loss": 0.3513, + "step": 2075 + }, + { + "epoch": 0.9082969432314411, + "grad_norm": 0.4054137512768172, + "learning_rate": 4.400334151617171e-05, + "loss": 0.3551, + "step": 2080 + }, + { + "epoch": 0.9104803493449781, + "grad_norm": 0.5160259798496531, + "learning_rate": 4.39620117889212e-05, + "loss": 0.3202, + "step": 2085 + }, + { + "epoch": 0.9126637554585153, + "grad_norm": 1.180298141482782, + "learning_rate": 4.3920559677446335e-05, + "loss": 0.3264, + "step": 2090 + }, + { + "epoch": 0.9148471615720524, + "grad_norm": 0.4807710172298594, + "learning_rate": 4.387898544928648e-05, + "loss": 0.3382, + "step": 2095 + }, + { + "epoch": 0.9170305676855895, + "grad_norm": 0.5325040390840762, + "learning_rate": 4.383728937276915e-05, + "loss": 0.3408, + "step": 2100 + }, + { + "epoch": 0.9192139737991266, + "grad_norm": 0.4423809835776037, + "learning_rate": 4.37954717170083e-05, + "loss": 0.3336, + "step": 2105 + }, + { + "epoch": 0.9213973799126638, + "grad_norm": 0.5498410601529987, + "learning_rate": 4.375353275190259e-05, + "loss": 0.3309, + "step": 2110 + }, + { + "epoch": 0.9235807860262009, + "grad_norm": 0.37076013082197373, + "learning_rate": 4.371147274813362e-05, + "loss": 0.3414, + "step": 2115 + }, + { + "epoch": 0.925764192139738, + "grad_norm": 0.403897556358072, + "learning_rate": 4.3669291977164185e-05, + "loss": 0.3308, + "step": 2120 + }, + { + "epoch": 0.9279475982532751, + "grad_norm": 0.4050123744090438, + "learning_rate": 4.362699071123655e-05, + "loss": 0.3383, + "step": 2125 + }, + { + "epoch": 0.9301310043668122, + "grad_norm": 0.4499616242086888, + "learning_rate": 4.358456922337068e-05, + "loss": 0.3247, + "step": 2130 + }, + { + "epoch": 0.9323144104803494, + "grad_norm": 0.41076931445914805, + "learning_rate": 4.354202778736245e-05, + "loss": 0.3325, + "step": 2135 + }, + { + "epoch": 0.9344978165938864, + "grad_norm": 0.6474266802788862, + "learning_rate": 4.3499366677781927e-05, + "loss": 0.3384, + "step": 2140 + }, + { + "epoch": 0.9366812227074236, + "grad_norm": 0.43958950888079396, + "learning_rate": 4.3456586169971555e-05, + "loss": 0.3155, + "step": 2145 + }, + { + "epoch": 0.9388646288209607, + "grad_norm": 0.430386825850428, + "learning_rate": 4.341368654004442e-05, + "loss": 0.3407, + "step": 2150 + }, + { + "epoch": 0.9410480349344978, + "grad_norm": 0.4400594923102947, + "learning_rate": 4.3370668064882406e-05, + "loss": 0.3227, + "step": 2155 + }, + { + "epoch": 0.9432314410480349, + "grad_norm": 0.49619465693365633, + "learning_rate": 4.332753102213447e-05, + "loss": 0.3488, + "step": 2160 + }, + { + "epoch": 0.9454148471615721, + "grad_norm": 0.2953653819485504, + "learning_rate": 4.328427569021481e-05, + "loss": 0.3237, + "step": 2165 + }, + { + "epoch": 0.9475982532751092, + "grad_norm": 0.3277990570496694, + "learning_rate": 4.3240902348301134e-05, + "loss": 0.3137, + "step": 2170 + }, + { + "epoch": 0.9497816593886463, + "grad_norm": 0.31994132171967327, + "learning_rate": 4.319741127633273e-05, + "loss": 0.3263, + "step": 2175 + }, + { + "epoch": 0.9519650655021834, + "grad_norm": 0.3159818487824089, + "learning_rate": 4.315380275500879e-05, + "loss": 0.3342, + "step": 2180 + }, + { + "epoch": 0.9541484716157205, + "grad_norm": 0.37710542197615304, + "learning_rate": 4.311007706578654e-05, + "loss": 0.3202, + "step": 2185 + }, + { + "epoch": 0.9563318777292577, + "grad_norm": 0.39548613497208596, + "learning_rate": 4.306623449087939e-05, + "loss": 0.3182, + "step": 2190 + }, + { + "epoch": 0.9585152838427947, + "grad_norm": 0.3737799555086228, + "learning_rate": 4.3022275313255215e-05, + "loss": 0.3202, + "step": 2195 + }, + { + "epoch": 0.9606986899563319, + "grad_norm": 0.3302761729103403, + "learning_rate": 4.2978199816634404e-05, + "loss": 0.3134, + "step": 2200 + }, + { + "epoch": 0.962882096069869, + "grad_norm": 0.3169568046490519, + "learning_rate": 4.293400828548812e-05, + "loss": 0.3281, + "step": 2205 + }, + { + "epoch": 0.9650655021834061, + "grad_norm": 0.3469828187440111, + "learning_rate": 4.2889701005036444e-05, + "loss": 0.3257, + "step": 2210 + }, + { + "epoch": 0.9672489082969432, + "grad_norm": 0.41804346589863545, + "learning_rate": 4.28452782612465e-05, + "loss": 0.3395, + "step": 2215 + }, + { + "epoch": 0.9694323144104804, + "grad_norm": 0.33364679303892797, + "learning_rate": 4.2800740340830636e-05, + "loss": 0.3127, + "step": 2220 + }, + { + "epoch": 0.9716157205240175, + "grad_norm": 0.3192103795670632, + "learning_rate": 4.275608753124459e-05, + "loss": 0.3139, + "step": 2225 + }, + { + "epoch": 0.9737991266375546, + "grad_norm": 0.43164296461188517, + "learning_rate": 4.271132012068559e-05, + "loss": 0.3144, + "step": 2230 + }, + { + "epoch": 0.9759825327510917, + "grad_norm": 0.36786016009963335, + "learning_rate": 4.266643839809054e-05, + "loss": 0.3217, + "step": 2235 + }, + { + "epoch": 0.9781659388646288, + "grad_norm": 0.40515456026599855, + "learning_rate": 4.262144265313412e-05, + "loss": 0.3263, + "step": 2240 + }, + { + "epoch": 0.980349344978166, + "grad_norm": 0.40316039408536825, + "learning_rate": 4.2576333176226944e-05, + "loss": 0.3174, + "step": 2245 + }, + { + "epoch": 0.982532751091703, + "grad_norm": 0.4036377506232831, + "learning_rate": 4.253111025851366e-05, + "loss": 0.3324, + "step": 2250 + }, + { + "epoch": 0.9847161572052402, + "grad_norm": 0.38600938504920373, + "learning_rate": 4.2485774191871084e-05, + "loss": 0.3148, + "step": 2255 + }, + { + "epoch": 0.9868995633187773, + "grad_norm": 0.3724866092235095, + "learning_rate": 4.244032526890633e-05, + "loss": 0.32, + "step": 2260 + }, + { + "epoch": 0.9890829694323144, + "grad_norm": 0.45130824685960164, + "learning_rate": 4.239476378295489e-05, + "loss": 0.3208, + "step": 2265 + }, + { + "epoch": 0.9912663755458515, + "grad_norm": 0.3614299023632802, + "learning_rate": 4.2349090028078766e-05, + "loss": 0.3108, + "step": 2270 + }, + { + "epoch": 0.9934497816593887, + "grad_norm": 0.3529974003854878, + "learning_rate": 4.230330429906456e-05, + "loss": 0.3114, + "step": 2275 + }, + { + "epoch": 0.9956331877729258, + "grad_norm": 0.3358112816928416, + "learning_rate": 4.225740689142158e-05, + "loss": 0.3168, + "step": 2280 + }, + { + "epoch": 0.9978165938864629, + "grad_norm": 0.3907341561097172, + "learning_rate": 4.221139810137992e-05, + "loss": 0.3174, + "step": 2285 + }, + { + "epoch": 1.0, + "grad_norm": 0.37377145244747273, + "learning_rate": 4.216527822588857e-05, + "loss": 0.3168, + "step": 2290 + }, + { + "epoch": 1.002183406113537, + "grad_norm": 0.40852537721961585, + "learning_rate": 4.2119047562613476e-05, + "loss": 0.2941, + "step": 2295 + }, + { + "epoch": 1.0043668122270741, + "grad_norm": 0.39878675388447493, + "learning_rate": 4.207270640993561e-05, + "loss": 0.2787, + "step": 2300 + }, + { + "epoch": 1.0065502183406114, + "grad_norm": 0.346225297182408, + "learning_rate": 4.202625506694911e-05, + "loss": 0.2686, + "step": 2305 + }, + { + "epoch": 1.0087336244541485, + "grad_norm": 0.34208833101355196, + "learning_rate": 4.197969383345925e-05, + "loss": 0.2809, + "step": 2310 + }, + { + "epoch": 1.0109170305676856, + "grad_norm": 0.4002016670275176, + "learning_rate": 4.19330230099806e-05, + "loss": 0.2979, + "step": 2315 + }, + { + "epoch": 1.0131004366812226, + "grad_norm": 0.4630743803988168, + "learning_rate": 4.188624289773501e-05, + "loss": 0.2967, + "step": 2320 + }, + { + "epoch": 1.01528384279476, + "grad_norm": 0.3568696275266243, + "learning_rate": 4.1839353798649717e-05, + "loss": 0.2758, + "step": 2325 + }, + { + "epoch": 1.017467248908297, + "grad_norm": 0.3119249400022314, + "learning_rate": 4.1792356015355375e-05, + "loss": 0.2766, + "step": 2330 + }, + { + "epoch": 1.019650655021834, + "grad_norm": 0.3696696936235827, + "learning_rate": 4.17452498511841e-05, + "loss": 0.286, + "step": 2335 + }, + { + "epoch": 1.0218340611353711, + "grad_norm": 0.3084313639844132, + "learning_rate": 4.169803561016754e-05, + "loss": 0.2856, + "step": 2340 + }, + { + "epoch": 1.0240174672489082, + "grad_norm": 0.3909495635210935, + "learning_rate": 4.1650713597034845e-05, + "loss": 0.287, + "step": 2345 + }, + { + "epoch": 1.0262008733624455, + "grad_norm": 0.44840025564529157, + "learning_rate": 4.16032841172108e-05, + "loss": 0.2927, + "step": 2350 + }, + { + "epoch": 1.0283842794759825, + "grad_norm": 0.41340997676020685, + "learning_rate": 4.155574747681375e-05, + "loss": 0.2858, + "step": 2355 + }, + { + "epoch": 1.0305676855895196, + "grad_norm": 0.3994213300123044, + "learning_rate": 4.1508103982653715e-05, + "loss": 0.2991, + "step": 2360 + }, + { + "epoch": 1.0327510917030567, + "grad_norm": 0.33838520158365676, + "learning_rate": 4.146035394223034e-05, + "loss": 0.2776, + "step": 2365 + }, + { + "epoch": 1.034934497816594, + "grad_norm": 0.3368322156288848, + "learning_rate": 4.1412497663730954e-05, + "loss": 0.2748, + "step": 2370 + }, + { + "epoch": 1.037117903930131, + "grad_norm": 0.3141352669129108, + "learning_rate": 4.1364535456028555e-05, + "loss": 0.2841, + "step": 2375 + }, + { + "epoch": 1.039301310043668, + "grad_norm": 0.2818967412459555, + "learning_rate": 4.1316467628679834e-05, + "loss": 0.2619, + "step": 2380 + }, + { + "epoch": 1.0414847161572052, + "grad_norm": 0.28565982519466604, + "learning_rate": 4.126829449192316e-05, + "loss": 0.2857, + "step": 2385 + }, + { + "epoch": 1.0436681222707425, + "grad_norm": 0.3210535970776655, + "learning_rate": 4.1220016356676615e-05, + "loss": 0.2787, + "step": 2390 + }, + { + "epoch": 1.0458515283842795, + "grad_norm": 0.3919954733516357, + "learning_rate": 4.117163353453593e-05, + "loss": 0.2824, + "step": 2395 + }, + { + "epoch": 1.0480349344978166, + "grad_norm": 0.2781986797396297, + "learning_rate": 4.112314633777251e-05, + "loss": 0.2831, + "step": 2400 + }, + { + "epoch": 1.0502183406113537, + "grad_norm": 0.3117143962813532, + "learning_rate": 4.107455507933144e-05, + "loss": 0.2741, + "step": 2405 + }, + { + "epoch": 1.0524017467248907, + "grad_norm": 0.29749854620448457, + "learning_rate": 4.102586007282938e-05, + "loss": 0.2784, + "step": 2410 + }, + { + "epoch": 1.054585152838428, + "grad_norm": 0.3011787067490516, + "learning_rate": 4.0977061632552656e-05, + "loss": 0.2822, + "step": 2415 + }, + { + "epoch": 1.056768558951965, + "grad_norm": 0.27733319238652004, + "learning_rate": 4.092816007345515e-05, + "loss": 0.2918, + "step": 2420 + }, + { + "epoch": 1.0589519650655022, + "grad_norm": 0.3178714706734469, + "learning_rate": 4.087915571115629e-05, + "loss": 0.2657, + "step": 2425 + }, + { + "epoch": 1.0611353711790392, + "grad_norm": 0.33147612300712626, + "learning_rate": 4.083004886193902e-05, + "loss": 0.2821, + "step": 2430 + }, + { + "epoch": 1.0633187772925765, + "grad_norm": 0.32015236720081114, + "learning_rate": 4.0780839842747754e-05, + "loss": 0.2876, + "step": 2435 + }, + { + "epoch": 1.0655021834061136, + "grad_norm": 0.2637393126437815, + "learning_rate": 4.073152897118632e-05, + "loss": 0.2902, + "step": 2440 + }, + { + "epoch": 1.0676855895196506, + "grad_norm": 0.2971292010190096, + "learning_rate": 4.0682116565515935e-05, + "loss": 0.294, + "step": 2445 + }, + { + "epoch": 1.0698689956331877, + "grad_norm": 0.32151195135466043, + "learning_rate": 4.063260294465311e-05, + "loss": 0.2754, + "step": 2450 + }, + { + "epoch": 1.072052401746725, + "grad_norm": 0.29344847997392354, + "learning_rate": 4.058298842816765e-05, + "loss": 0.2756, + "step": 2455 + }, + { + "epoch": 1.074235807860262, + "grad_norm": 0.38141464593456376, + "learning_rate": 4.0533273336280516e-05, + "loss": 0.2749, + "step": 2460 + }, + { + "epoch": 1.0764192139737991, + "grad_norm": 0.3519071752481233, + "learning_rate": 4.048345798986186e-05, + "loss": 0.2937, + "step": 2465 + }, + { + "epoch": 1.0786026200873362, + "grad_norm": 0.27974724944596935, + "learning_rate": 4.043354271042884e-05, + "loss": 0.2772, + "step": 2470 + }, + { + "epoch": 1.0807860262008733, + "grad_norm": 0.31883200998020006, + "learning_rate": 4.038352782014362e-05, + "loss": 0.2783, + "step": 2475 + }, + { + "epoch": 1.0829694323144106, + "grad_norm": 0.3107436150415206, + "learning_rate": 4.033341364181128e-05, + "loss": 0.2845, + "step": 2480 + }, + { + "epoch": 1.0851528384279476, + "grad_norm": 0.3631114446623769, + "learning_rate": 4.028320049887771e-05, + "loss": 0.2815, + "step": 2485 + }, + { + "epoch": 1.0873362445414847, + "grad_norm": 0.29440410304916975, + "learning_rate": 4.023288871542752e-05, + "loss": 0.2767, + "step": 2490 + }, + { + "epoch": 1.0895196506550218, + "grad_norm": 0.3013728586687223, + "learning_rate": 4.0182478616182015e-05, + "loss": 0.2802, + "step": 2495 + }, + { + "epoch": 1.091703056768559, + "grad_norm": 0.30015696374357026, + "learning_rate": 4.013197052649699e-05, + "loss": 0.2885, + "step": 2500 + }, + { + "epoch": 1.0938864628820961, + "grad_norm": 0.334054468972433, + "learning_rate": 4.0081364772360725e-05, + "loss": 0.2849, + "step": 2505 + }, + { + "epoch": 1.0960698689956332, + "grad_norm": 0.29580484838617715, + "learning_rate": 4.003066168039183e-05, + "loss": 0.2748, + "step": 2510 + }, + { + "epoch": 1.0982532751091703, + "grad_norm": 0.3454428161947495, + "learning_rate": 3.9979861577837155e-05, + "loss": 0.2783, + "step": 2515 + }, + { + "epoch": 1.1004366812227073, + "grad_norm": 0.3354653145039301, + "learning_rate": 3.9928964792569655e-05, + "loss": 0.2811, + "step": 2520 + }, + { + "epoch": 1.1026200873362446, + "grad_norm": 0.3257148621799395, + "learning_rate": 3.987797165308633e-05, + "loss": 0.2805, + "step": 2525 + }, + { + "epoch": 1.1048034934497817, + "grad_norm": 0.3093405188202137, + "learning_rate": 3.982688248850604e-05, + "loss": 0.2929, + "step": 2530 + }, + { + "epoch": 1.1069868995633187, + "grad_norm": 0.42822993107757057, + "learning_rate": 3.977569762856741e-05, + "loss": 0.2913, + "step": 2535 + }, + { + "epoch": 1.1091703056768558, + "grad_norm": 0.40323824571841865, + "learning_rate": 3.9724417403626705e-05, + "loss": 0.2834, + "step": 2540 + }, + { + "epoch": 1.111353711790393, + "grad_norm": 0.30813240614187476, + "learning_rate": 3.9673042144655697e-05, + "loss": 0.2927, + "step": 2545 + }, + { + "epoch": 1.1135371179039302, + "grad_norm": 0.2807299601361123, + "learning_rate": 3.9621572183239515e-05, + "loss": 0.2778, + "step": 2550 + }, + { + "epoch": 1.1157205240174672, + "grad_norm": 0.34539465437923245, + "learning_rate": 3.957000785157452e-05, + "loss": 0.3014, + "step": 2555 + }, + { + "epoch": 1.1179039301310043, + "grad_norm": 0.36001154980798994, + "learning_rate": 3.951834948246616e-05, + "loss": 0.268, + "step": 2560 + }, + { + "epoch": 1.1200873362445414, + "grad_norm": 0.3659504200165266, + "learning_rate": 3.946659740932681e-05, + "loss": 0.2999, + "step": 2565 + }, + { + "epoch": 1.1222707423580787, + "grad_norm": 0.310758158636425, + "learning_rate": 3.9414751966173634e-05, + "loss": 0.2788, + "step": 2570 + }, + { + "epoch": 1.1244541484716157, + "grad_norm": 0.2658948188578196, + "learning_rate": 3.93628134876264e-05, + "loss": 0.2764, + "step": 2575 + }, + { + "epoch": 1.1266375545851528, + "grad_norm": 0.3038259778170334, + "learning_rate": 3.9310782308905395e-05, + "loss": 0.2809, + "step": 2580 + }, + { + "epoch": 1.1288209606986899, + "grad_norm": 0.3098003775343629, + "learning_rate": 3.925865876582915e-05, + "loss": 0.2668, + "step": 2585 + }, + { + "epoch": 1.1310043668122272, + "grad_norm": 0.29478313929574995, + "learning_rate": 3.920644319481237e-05, + "loss": 0.2762, + "step": 2590 + }, + { + "epoch": 1.1331877729257642, + "grad_norm": 0.2715517904632468, + "learning_rate": 3.915413593286371e-05, + "loss": 0.2828, + "step": 2595 + }, + { + "epoch": 1.1353711790393013, + "grad_norm": 0.2546601742913032, + "learning_rate": 3.910173731758362e-05, + "loss": 0.2752, + "step": 2600 + }, + { + "epoch": 1.1375545851528384, + "grad_norm": 0.3020454024382658, + "learning_rate": 3.904924768716216e-05, + "loss": 0.2764, + "step": 2605 + }, + { + "epoch": 1.1397379912663754, + "grad_norm": 0.3404890082601122, + "learning_rate": 3.89966673803768e-05, + "loss": 0.2746, + "step": 2610 + }, + { + "epoch": 1.1419213973799127, + "grad_norm": 0.2918728991626841, + "learning_rate": 3.894399673659031e-05, + "loss": 0.2911, + "step": 2615 + }, + { + "epoch": 1.1441048034934498, + "grad_norm": 0.32777823728250605, + "learning_rate": 3.889123609574843e-05, + "loss": 0.2687, + "step": 2620 + }, + { + "epoch": 1.1462882096069869, + "grad_norm": 0.3417220224406648, + "learning_rate": 3.883838579837781e-05, + "loss": 0.29, + "step": 2625 + }, + { + "epoch": 1.1484716157205241, + "grad_norm": 0.3206470787242044, + "learning_rate": 3.8785446185583755e-05, + "loss": 0.2852, + "step": 2630 + }, + { + "epoch": 1.1506550218340612, + "grad_norm": 0.2896917429374818, + "learning_rate": 3.873241759904801e-05, + "loss": 0.2844, + "step": 2635 + }, + { + "epoch": 1.1528384279475983, + "grad_norm": 0.319012549688518, + "learning_rate": 3.8679300381026585e-05, + "loss": 0.2926, + "step": 2640 + }, + { + "epoch": 1.1550218340611353, + "grad_norm": 0.27491269743019364, + "learning_rate": 3.862609487434752e-05, + "loss": 0.2701, + "step": 2645 + }, + { + "epoch": 1.1572052401746724, + "grad_norm": 0.30736358656559387, + "learning_rate": 3.857280142240871e-05, + "loss": 0.2813, + "step": 2650 + }, + { + "epoch": 1.1593886462882097, + "grad_norm": 0.27847789277867213, + "learning_rate": 3.851942036917565e-05, + "loss": 0.2989, + "step": 2655 + }, + { + "epoch": 1.1615720524017468, + "grad_norm": 0.3178443314858301, + "learning_rate": 3.846595205917922e-05, + "loss": 0.2831, + "step": 2660 + }, + { + "epoch": 1.1637554585152838, + "grad_norm": 0.27407489870985063, + "learning_rate": 3.8412396837513484e-05, + "loss": 0.2838, + "step": 2665 + }, + { + "epoch": 1.165938864628821, + "grad_norm": 0.31079163261067777, + "learning_rate": 3.835875504983344e-05, + "loss": 0.2896, + "step": 2670 + }, + { + "epoch": 1.1681222707423582, + "grad_norm": 0.330388527773976, + "learning_rate": 3.830502704235282e-05, + "loss": 0.281, + "step": 2675 + }, + { + "epoch": 1.1703056768558953, + "grad_norm": 0.28816900411883056, + "learning_rate": 3.825121316184181e-05, + "loss": 0.2883, + "step": 2680 + }, + { + "epoch": 1.1724890829694323, + "grad_norm": 0.3328938276296137, + "learning_rate": 3.8197313755624855e-05, + "loss": 0.266, + "step": 2685 + }, + { + "epoch": 1.1746724890829694, + "grad_norm": 0.3824721867696784, + "learning_rate": 3.8143329171578366e-05, + "loss": 0.2792, + "step": 2690 + }, + { + "epoch": 1.1768558951965065, + "grad_norm": 0.33578112498653234, + "learning_rate": 3.808925975812855e-05, + "loss": 0.2994, + "step": 2695 + }, + { + "epoch": 1.1790393013100438, + "grad_norm": 0.35312074931486553, + "learning_rate": 3.803510586424908e-05, + "loss": 0.2909, + "step": 2700 + }, + { + "epoch": 1.1812227074235808, + "grad_norm": 0.3016960624575352, + "learning_rate": 3.798086783945892e-05, + "loss": 0.2829, + "step": 2705 + }, + { + "epoch": 1.1834061135371179, + "grad_norm": 0.355359254971825, + "learning_rate": 3.792654603381998e-05, + "loss": 0.2843, + "step": 2710 + }, + { + "epoch": 1.185589519650655, + "grad_norm": 0.3700674234618425, + "learning_rate": 3.7872140797934956e-05, + "loss": 0.2719, + "step": 2715 + }, + { + "epoch": 1.1877729257641922, + "grad_norm": 0.44056476547643436, + "learning_rate": 3.781765248294499e-05, + "loss": 0.2759, + "step": 2720 + }, + { + "epoch": 1.1899563318777293, + "grad_norm": 0.29079420170197845, + "learning_rate": 3.7763081440527435e-05, + "loss": 0.2857, + "step": 2725 + }, + { + "epoch": 1.1921397379912664, + "grad_norm": 0.312707282955329, + "learning_rate": 3.7708428022893596e-05, + "loss": 0.2863, + "step": 2730 + }, + { + "epoch": 1.1943231441048034, + "grad_norm": 0.2684008615845062, + "learning_rate": 3.7653692582786416e-05, + "loss": 0.272, + "step": 2735 + }, + { + "epoch": 1.1965065502183405, + "grad_norm": 0.35782293675245397, + "learning_rate": 3.7598875473478246e-05, + "loss": 0.2715, + "step": 2740 + }, + { + "epoch": 1.1986899563318778, + "grad_norm": 0.30301767808558216, + "learning_rate": 3.7543977048768534e-05, + "loss": 0.271, + "step": 2745 + }, + { + "epoch": 1.2008733624454149, + "grad_norm": 0.3151321773465957, + "learning_rate": 3.748899766298157e-05, + "loss": 0.2845, + "step": 2750 + }, + { + "epoch": 1.203056768558952, + "grad_norm": 0.2925859858510235, + "learning_rate": 3.743393767096416e-05, + "loss": 0.2755, + "step": 2755 + }, + { + "epoch": 1.205240174672489, + "grad_norm": 0.2859805443710802, + "learning_rate": 3.737879742808337e-05, + "loss": 0.2793, + "step": 2760 + }, + { + "epoch": 1.2074235807860263, + "grad_norm": 0.25461387164265176, + "learning_rate": 3.732357729022421e-05, + "loss": 0.2877, + "step": 2765 + }, + { + "epoch": 1.2096069868995634, + "grad_norm": 0.24813193774999823, + "learning_rate": 3.7268277613787356e-05, + "loss": 0.2869, + "step": 2770 + }, + { + "epoch": 1.2117903930131004, + "grad_norm": 0.3276586080798546, + "learning_rate": 3.721289875568683e-05, + "loss": 0.2853, + "step": 2775 + }, + { + "epoch": 1.2139737991266375, + "grad_norm": 0.3351580255362412, + "learning_rate": 3.715744107334773e-05, + "loss": 0.2699, + "step": 2780 + }, + { + "epoch": 1.2161572052401746, + "grad_norm": 0.30882004844208494, + "learning_rate": 3.7101904924703865e-05, + "loss": 0.2823, + "step": 2785 + }, + { + "epoch": 1.2183406113537119, + "grad_norm": 0.37760317246667774, + "learning_rate": 3.704629066819549e-05, + "loss": 0.2771, + "step": 2790 + }, + { + "epoch": 1.220524017467249, + "grad_norm": 0.32894843491571224, + "learning_rate": 3.6990598662767014e-05, + "loss": 0.2699, + "step": 2795 + }, + { + "epoch": 1.222707423580786, + "grad_norm": 0.3476225470830814, + "learning_rate": 3.693482926786461e-05, + "loss": 0.2726, + "step": 2800 + }, + { + "epoch": 1.224890829694323, + "grad_norm": 0.3118303300626587, + "learning_rate": 3.687898284343396e-05, + "loss": 0.2891, + "step": 2805 + }, + { + "epoch": 1.2270742358078603, + "grad_norm": 0.3281515137713933, + "learning_rate": 3.68230597499179e-05, + "loss": 0.2805, + "step": 2810 + }, + { + "epoch": 1.2292576419213974, + "grad_norm": 0.281506457906696, + "learning_rate": 3.676706034825409e-05, + "loss": 0.3019, + "step": 2815 + }, + { + "epoch": 1.2314410480349345, + "grad_norm": 0.2918564930941491, + "learning_rate": 3.6710984999872735e-05, + "loss": 0.2748, + "step": 2820 + }, + { + "epoch": 1.2336244541484715, + "grad_norm": 0.301158978005547, + "learning_rate": 3.665483406669418e-05, + "loss": 0.2904, + "step": 2825 + }, + { + "epoch": 1.2358078602620086, + "grad_norm": 0.29117932725379103, + "learning_rate": 3.6598607911126614e-05, + "loss": 0.2856, + "step": 2830 + }, + { + "epoch": 1.237991266375546, + "grad_norm": 0.2736978249597251, + "learning_rate": 3.654230689606372e-05, + "loss": 0.2659, + "step": 2835 + }, + { + "epoch": 1.240174672489083, + "grad_norm": 0.2595195793059447, + "learning_rate": 3.648593138488235e-05, + "loss": 0.278, + "step": 2840 + }, + { + "epoch": 1.24235807860262, + "grad_norm": 0.31138011895796025, + "learning_rate": 3.6429481741440176e-05, + "loss": 0.2804, + "step": 2845 + }, + { + "epoch": 1.244541484716157, + "grad_norm": 0.28491002649715325, + "learning_rate": 3.63729583300733e-05, + "loss": 0.278, + "step": 2850 + }, + { + "epoch": 1.2467248908296944, + "grad_norm": 0.28329885485362183, + "learning_rate": 3.6316361515593966e-05, + "loss": 0.2874, + "step": 2855 + }, + { + "epoch": 1.2489082969432315, + "grad_norm": 0.36368480663196673, + "learning_rate": 3.625969166328815e-05, + "loss": 0.276, + "step": 2860 + }, + { + "epoch": 1.2510917030567685, + "grad_norm": 0.2700235560170439, + "learning_rate": 3.620294913891326e-05, + "loss": 0.2884, + "step": 2865 + }, + { + "epoch": 1.2532751091703056, + "grad_norm": 0.25418847496227326, + "learning_rate": 3.614613430869571e-05, + "loss": 0.2859, + "step": 2870 + }, + { + "epoch": 1.2554585152838427, + "grad_norm": 0.28230477688636335, + "learning_rate": 3.608924753932862e-05, + "loss": 0.2884, + "step": 2875 + }, + { + "epoch": 1.25764192139738, + "grad_norm": 0.31064202314287603, + "learning_rate": 3.6032289197969374e-05, + "loss": 0.2858, + "step": 2880 + }, + { + "epoch": 1.259825327510917, + "grad_norm": 0.272199049103352, + "learning_rate": 3.5975259652237333e-05, + "loss": 0.287, + "step": 2885 + }, + { + "epoch": 1.262008733624454, + "grad_norm": 0.23256991109619604, + "learning_rate": 3.5918159270211424e-05, + "loss": 0.2734, + "step": 2890 + }, + { + "epoch": 1.2641921397379914, + "grad_norm": 0.2926902120837456, + "learning_rate": 3.586098842042773e-05, + "loss": 0.2774, + "step": 2895 + }, + { + "epoch": 1.2663755458515285, + "grad_norm": 0.36624779554626996, + "learning_rate": 3.580374747187719e-05, + "loss": 0.2775, + "step": 2900 + }, + { + "epoch": 1.2685589519650655, + "grad_norm": 0.29331268838279223, + "learning_rate": 3.574643679400313e-05, + "loss": 0.2747, + "step": 2905 + }, + { + "epoch": 1.2707423580786026, + "grad_norm": 0.27621174524527253, + "learning_rate": 3.568905675669895e-05, + "loss": 0.2882, + "step": 2910 + }, + { + "epoch": 1.2729257641921397, + "grad_norm": 0.2771703632616136, + "learning_rate": 3.5631607730305694e-05, + "loss": 0.2664, + "step": 2915 + }, + { + "epoch": 1.2751091703056767, + "grad_norm": 0.33588301272251375, + "learning_rate": 3.5574090085609696e-05, + "loss": 0.2798, + "step": 2920 + }, + { + "epoch": 1.277292576419214, + "grad_norm": 0.35009121663132425, + "learning_rate": 3.551650419384013e-05, + "loss": 0.2833, + "step": 2925 + }, + { + "epoch": 1.279475982532751, + "grad_norm": 0.2687807965233062, + "learning_rate": 3.545885042666669e-05, + "loss": 0.2738, + "step": 2930 + }, + { + "epoch": 1.2816593886462881, + "grad_norm": 0.36331931036035, + "learning_rate": 3.5401129156197105e-05, + "loss": 0.2833, + "step": 2935 + }, + { + "epoch": 1.2838427947598254, + "grad_norm": 0.3828328306666377, + "learning_rate": 3.534334075497482e-05, + "loss": 0.2821, + "step": 2940 + }, + { + "epoch": 1.2860262008733625, + "grad_norm": 0.30473362904993256, + "learning_rate": 3.5285485595976534e-05, + "loss": 0.2821, + "step": 2945 + }, + { + "epoch": 1.2882096069868996, + "grad_norm": 0.3499861198156561, + "learning_rate": 3.5227564052609816e-05, + "loss": 0.2724, + "step": 2950 + }, + { + "epoch": 1.2903930131004366, + "grad_norm": 0.2938833486098817, + "learning_rate": 3.5169576498710696e-05, + "loss": 0.278, + "step": 2955 + }, + { + "epoch": 1.2925764192139737, + "grad_norm": 0.28084189095150464, + "learning_rate": 3.511152330854125e-05, + "loss": 0.2679, + "step": 2960 + }, + { + "epoch": 1.294759825327511, + "grad_norm": 0.3199421686229126, + "learning_rate": 3.505340485678717e-05, + "loss": 0.2819, + "step": 2965 + }, + { + "epoch": 1.296943231441048, + "grad_norm": 0.3149994117491474, + "learning_rate": 3.499522151855537e-05, + "loss": 0.2975, + "step": 2970 + }, + { + "epoch": 1.2991266375545851, + "grad_norm": 0.2899693019311649, + "learning_rate": 3.493697366937154e-05, + "loss": 0.2706, + "step": 2975 + }, + { + "epoch": 1.3013100436681222, + "grad_norm": 0.33913499909626604, + "learning_rate": 3.487866168517775e-05, + "loss": 0.294, + "step": 2980 + }, + { + "epoch": 1.3034934497816595, + "grad_norm": 0.3769147657930795, + "learning_rate": 3.4820285942330003e-05, + "loss": 0.2818, + "step": 2985 + }, + { + "epoch": 1.3056768558951966, + "grad_norm": 0.32801615796861155, + "learning_rate": 3.476184681759581e-05, + "loss": 0.28, + "step": 2990 + }, + { + "epoch": 1.3078602620087336, + "grad_norm": 0.27357707367933753, + "learning_rate": 3.4703344688151776e-05, + "loss": 0.2783, + "step": 2995 + }, + { + "epoch": 1.3100436681222707, + "grad_norm": 0.37622537438391995, + "learning_rate": 3.46447799315811e-05, + "loss": 0.2819, + "step": 3000 + }, + { + "epoch": 1.3122270742358078, + "grad_norm": 0.3550543812495112, + "learning_rate": 3.458615292587125e-05, + "loss": 0.2792, + "step": 3005 + }, + { + "epoch": 1.314410480349345, + "grad_norm": 0.2816496834529316, + "learning_rate": 3.452746404941143e-05, + "loss": 0.2859, + "step": 3010 + }, + { + "epoch": 1.3165938864628821, + "grad_norm": 0.3066581463920424, + "learning_rate": 3.446871368099017e-05, + "loss": 0.2746, + "step": 3015 + }, + { + "epoch": 1.3187772925764192, + "grad_norm": 0.2708116400301654, + "learning_rate": 3.440990219979289e-05, + "loss": 0.2863, + "step": 3020 + }, + { + "epoch": 1.3209606986899565, + "grad_norm": 0.2497661453451677, + "learning_rate": 3.435102998539942e-05, + "loss": 0.2702, + "step": 3025 + }, + { + "epoch": 1.3231441048034935, + "grad_norm": 0.2651767333352664, + "learning_rate": 3.4292097417781595e-05, + "loss": 0.2695, + "step": 3030 + }, + { + "epoch": 1.3253275109170306, + "grad_norm": 0.2695245871370125, + "learning_rate": 3.4233104877300775e-05, + "loss": 0.267, + "step": 3035 + }, + { + "epoch": 1.3275109170305677, + "grad_norm": 0.27077504903552757, + "learning_rate": 3.41740527447054e-05, + "loss": 0.2662, + "step": 3040 + }, + { + "epoch": 1.3296943231441047, + "grad_norm": 0.2519617614699693, + "learning_rate": 3.411494140112852e-05, + "loss": 0.2716, + "step": 3045 + }, + { + "epoch": 1.3318777292576418, + "grad_norm": 0.2809728902320245, + "learning_rate": 3.4055771228085347e-05, + "loss": 0.2879, + "step": 3050 + }, + { + "epoch": 1.334061135371179, + "grad_norm": 0.26727497601257166, + "learning_rate": 3.399654260747078e-05, + "loss": 0.2725, + "step": 3055 + }, + { + "epoch": 1.3362445414847162, + "grad_norm": 0.34198746633106286, + "learning_rate": 3.3937255921556974e-05, + "loss": 0.2877, + "step": 3060 + }, + { + "epoch": 1.3384279475982532, + "grad_norm": 0.2984040980470414, + "learning_rate": 3.3877911552990826e-05, + "loss": 0.2793, + "step": 3065 + }, + { + "epoch": 1.3406113537117905, + "grad_norm": 0.2865962833130215, + "learning_rate": 3.3818509884791514e-05, + "loss": 0.273, + "step": 3070 + }, + { + "epoch": 1.3427947598253276, + "grad_norm": 0.2988181873287166, + "learning_rate": 3.375905130034808e-05, + "loss": 0.275, + "step": 3075 + }, + { + "epoch": 1.3449781659388647, + "grad_norm": 0.29108551078775924, + "learning_rate": 3.369953618341688e-05, + "loss": 0.2874, + "step": 3080 + }, + { + "epoch": 1.3471615720524017, + "grad_norm": 0.29395564574021693, + "learning_rate": 3.3639964918119135e-05, + "loss": 0.2756, + "step": 3085 + }, + { + "epoch": 1.3493449781659388, + "grad_norm": 0.2834037344232317, + "learning_rate": 3.3580337888938515e-05, + "loss": 0.2795, + "step": 3090 + }, + { + "epoch": 1.3515283842794759, + "grad_norm": 0.2963729779205327, + "learning_rate": 3.3520655480718515e-05, + "loss": 0.2821, + "step": 3095 + }, + { + "epoch": 1.3537117903930131, + "grad_norm": 0.2719250597184558, + "learning_rate": 3.346091807866013e-05, + "loss": 0.2799, + "step": 3100 + }, + { + "epoch": 1.3558951965065502, + "grad_norm": 0.345480183079055, + "learning_rate": 3.3401126068319244e-05, + "loss": 0.2753, + "step": 3105 + }, + { + "epoch": 1.3580786026200873, + "grad_norm": 0.2731322103388952, + "learning_rate": 3.3341279835604224e-05, + "loss": 0.2806, + "step": 3110 + }, + { + "epoch": 1.3602620087336246, + "grad_norm": 0.3377724432178437, + "learning_rate": 3.328137976677339e-05, + "loss": 0.2952, + "step": 3115 + }, + { + "epoch": 1.3624454148471616, + "grad_norm": 0.3490702531757625, + "learning_rate": 3.322142624843253e-05, + "loss": 0.2883, + "step": 3120 + }, + { + "epoch": 1.3646288209606987, + "grad_norm": 0.23733767943987039, + "learning_rate": 3.3161419667532384e-05, + "loss": 0.2811, + "step": 3125 + }, + { + "epoch": 1.3668122270742358, + "grad_norm": 0.29640032968995195, + "learning_rate": 3.3101360411366205e-05, + "loss": 0.2714, + "step": 3130 + }, + { + "epoch": 1.3689956331877728, + "grad_norm": 0.38335436641971393, + "learning_rate": 3.304124886756719e-05, + "loss": 0.2773, + "step": 3135 + }, + { + "epoch": 1.37117903930131, + "grad_norm": 0.2614713678332863, + "learning_rate": 3.2981085424106036e-05, + "loss": 0.2716, + "step": 3140 + }, + { + "epoch": 1.3733624454148472, + "grad_norm": 0.271556464815005, + "learning_rate": 3.2920870469288375e-05, + "loss": 0.2929, + "step": 3145 + }, + { + "epoch": 1.3755458515283843, + "grad_norm": 0.2984513438432199, + "learning_rate": 3.2860604391752344e-05, + "loss": 0.2767, + "step": 3150 + }, + { + "epoch": 1.3777292576419213, + "grad_norm": 0.35226423449647926, + "learning_rate": 3.2800287580466e-05, + "loss": 0.2803, + "step": 3155 + }, + { + "epoch": 1.3799126637554586, + "grad_norm": 0.2762721435202866, + "learning_rate": 3.273992042472487e-05, + "loss": 0.2635, + "step": 3160 + }, + { + "epoch": 1.3820960698689957, + "grad_norm": 0.2849399596588635, + "learning_rate": 3.2679503314149395e-05, + "loss": 0.2828, + "step": 3165 + }, + { + "epoch": 1.3842794759825328, + "grad_norm": 0.3240823060545034, + "learning_rate": 3.261903663868243e-05, + "loss": 0.2772, + "step": 3170 + }, + { + "epoch": 1.3864628820960698, + "grad_norm": 0.29529588632218107, + "learning_rate": 3.2558520788586764e-05, + "loss": 0.2709, + "step": 3175 + }, + { + "epoch": 1.388646288209607, + "grad_norm": 0.2880598231126244, + "learning_rate": 3.249795615444253e-05, + "loss": 0.2649, + "step": 3180 + }, + { + "epoch": 1.3908296943231442, + "grad_norm": 0.34085771304801965, + "learning_rate": 3.243734312714475e-05, + "loss": 0.2767, + "step": 3185 + }, + { + "epoch": 1.3930131004366813, + "grad_norm": 0.24864813820746293, + "learning_rate": 3.2376682097900754e-05, + "loss": 0.2793, + "step": 3190 + }, + { + "epoch": 1.3951965065502183, + "grad_norm": 0.3059620331889871, + "learning_rate": 3.2315973458227714e-05, + "loss": 0.2875, + "step": 3195 + }, + { + "epoch": 1.3973799126637554, + "grad_norm": 0.24526372956599277, + "learning_rate": 3.225521759995007e-05, + "loss": 0.2821, + "step": 3200 + }, + { + "epoch": 1.3995633187772927, + "grad_norm": 0.2972629377196201, + "learning_rate": 3.219441491519703e-05, + "loss": 0.2742, + "step": 3205 + }, + { + "epoch": 1.4017467248908297, + "grad_norm": 0.30368222569014375, + "learning_rate": 3.2133565796400046e-05, + "loss": 0.2794, + "step": 3210 + }, + { + "epoch": 1.4039301310043668, + "grad_norm": 0.27087580419910157, + "learning_rate": 3.2072670636290195e-05, + "loss": 0.2686, + "step": 3215 + }, + { + "epoch": 1.4061135371179039, + "grad_norm": 0.3097415372011331, + "learning_rate": 3.20117298278958e-05, + "loss": 0.2744, + "step": 3220 + }, + { + "epoch": 1.408296943231441, + "grad_norm": 0.265200535542659, + "learning_rate": 3.195074376453975e-05, + "loss": 0.2904, + "step": 3225 + }, + { + "epoch": 1.4104803493449782, + "grad_norm": 0.26447053395871756, + "learning_rate": 3.188971283983703e-05, + "loss": 0.2633, + "step": 3230 + }, + { + "epoch": 1.4126637554585153, + "grad_norm": 0.25630242114154234, + "learning_rate": 3.1828637447692187e-05, + "loss": 0.2755, + "step": 3235 + }, + { + "epoch": 1.4148471615720524, + "grad_norm": 0.26664274806120797, + "learning_rate": 3.176751798229673e-05, + "loss": 0.2802, + "step": 3240 + }, + { + "epoch": 1.4170305676855894, + "grad_norm": 0.3278108102940346, + "learning_rate": 3.1706354838126687e-05, + "loss": 0.2585, + "step": 3245 + }, + { + "epoch": 1.4192139737991267, + "grad_norm": 0.24680968950816637, + "learning_rate": 3.1645148409939915e-05, + "loss": 0.2758, + "step": 3250 + }, + { + "epoch": 1.4213973799126638, + "grad_norm": 0.2843693481761553, + "learning_rate": 3.158389909277371e-05, + "loss": 0.2857, + "step": 3255 + }, + { + "epoch": 1.4235807860262009, + "grad_norm": 0.27926774884591915, + "learning_rate": 3.152260728194213e-05, + "loss": 0.2827, + "step": 3260 + }, + { + "epoch": 1.425764192139738, + "grad_norm": 0.33114453429263857, + "learning_rate": 3.1461273373033516e-05, + "loss": 0.2791, + "step": 3265 + }, + { + "epoch": 1.427947598253275, + "grad_norm": 0.2523111291701471, + "learning_rate": 3.1399897761907914e-05, + "loss": 0.2785, + "step": 3270 + }, + { + "epoch": 1.4301310043668123, + "grad_norm": 0.2669982957777147, + "learning_rate": 3.133848084469454e-05, + "loss": 0.2723, + "step": 3275 + }, + { + "epoch": 1.4323144104803494, + "grad_norm": 0.2551505338781039, + "learning_rate": 3.1277023017789164e-05, + "loss": 0.2701, + "step": 3280 + }, + { + "epoch": 1.4344978165938864, + "grad_norm": 0.2729328232441411, + "learning_rate": 3.121552467785164e-05, + "loss": 0.2756, + "step": 3285 + }, + { + "epoch": 1.4366812227074237, + "grad_norm": 0.31413945020452155, + "learning_rate": 3.1153986221803284e-05, + "loss": 0.2634, + "step": 3290 + }, + { + "epoch": 1.4388646288209608, + "grad_norm": 0.27459622465012024, + "learning_rate": 3.109240804682431e-05, + "loss": 0.2749, + "step": 3295 + }, + { + "epoch": 1.4410480349344978, + "grad_norm": 0.3428676752152493, + "learning_rate": 3.103079055035131e-05, + "loss": 0.2737, + "step": 3300 + }, + { + "epoch": 1.443231441048035, + "grad_norm": 0.28826863252341717, + "learning_rate": 3.096913413007467e-05, + "loss": 0.2716, + "step": 3305 + }, + { + "epoch": 1.445414847161572, + "grad_norm": 0.3003661345160089, + "learning_rate": 3.090743918393596e-05, + "loss": 0.2711, + "step": 3310 + }, + { + "epoch": 1.447598253275109, + "grad_norm": 0.2726421109267964, + "learning_rate": 3.0845706110125435e-05, + "loss": 0.2755, + "step": 3315 + }, + { + "epoch": 1.4497816593886463, + "grad_norm": 0.2633565745117208, + "learning_rate": 3.078393530707942e-05, + "loss": 0.2708, + "step": 3320 + }, + { + "epoch": 1.4519650655021834, + "grad_norm": 0.3590173494625973, + "learning_rate": 3.072212717347776e-05, + "loss": 0.2802, + "step": 3325 + }, + { + "epoch": 1.4541484716157205, + "grad_norm": 0.2841165489479175, + "learning_rate": 3.0660282108241235e-05, + "loss": 0.2793, + "step": 3330 + }, + { + "epoch": 1.4563318777292578, + "grad_norm": 0.27787135238996175, + "learning_rate": 3.059840051052898e-05, + "loss": 0.2915, + "step": 3335 + }, + { + "epoch": 1.4585152838427948, + "grad_norm": 0.29883394182548156, + "learning_rate": 3.053648277973592e-05, + "loss": 0.2667, + "step": 3340 + }, + { + "epoch": 1.460698689956332, + "grad_norm": 0.2897989278866621, + "learning_rate": 3.04745293154902e-05, + "loss": 0.2863, + "step": 3345 + }, + { + "epoch": 1.462882096069869, + "grad_norm": 0.2413426041774491, + "learning_rate": 3.0412540517650582e-05, + "loss": 0.2686, + "step": 3350 + }, + { + "epoch": 1.465065502183406, + "grad_norm": 0.3092426474858127, + "learning_rate": 3.03505167863039e-05, + "loss": 0.2861, + "step": 3355 + }, + { + "epoch": 1.467248908296943, + "grad_norm": 0.37271590450228287, + "learning_rate": 3.0288458521762414e-05, + "loss": 0.2804, + "step": 3360 + }, + { + "epoch": 1.4694323144104804, + "grad_norm": 0.32074172110600374, + "learning_rate": 3.02263661245613e-05, + "loss": 0.2673, + "step": 3365 + }, + { + "epoch": 1.4716157205240175, + "grad_norm": 0.26878027230807305, + "learning_rate": 3.016423999545603e-05, + "loss": 0.2805, + "step": 3370 + }, + { + "epoch": 1.4737991266375545, + "grad_norm": 0.2611222636406977, + "learning_rate": 3.0102080535419775e-05, + "loss": 0.2852, + "step": 3375 + }, + { + "epoch": 1.4759825327510918, + "grad_norm": 0.32711140245362424, + "learning_rate": 3.0039888145640845e-05, + "loss": 0.2768, + "step": 3380 + }, + { + "epoch": 1.4781659388646289, + "grad_norm": 0.2933942315221079, + "learning_rate": 2.9977663227520065e-05, + "loss": 0.2733, + "step": 3385 + }, + { + "epoch": 1.480349344978166, + "grad_norm": 0.3239911145447236, + "learning_rate": 2.991540618266822e-05, + "loss": 0.2689, + "step": 3390 + }, + { + "epoch": 1.482532751091703, + "grad_norm": 0.31053885049245833, + "learning_rate": 2.9853117412903443e-05, + "loss": 0.259, + "step": 3395 + }, + { + "epoch": 1.48471615720524, + "grad_norm": 0.29558451827845905, + "learning_rate": 2.979079732024862e-05, + "loss": 0.2676, + "step": 3400 + }, + { + "epoch": 1.4868995633187774, + "grad_norm": 0.2919836637875955, + "learning_rate": 2.9728446306928813e-05, + "loss": 0.274, + "step": 3405 + }, + { + "epoch": 1.4890829694323144, + "grad_norm": 0.2914986515005281, + "learning_rate": 2.966606477536863e-05, + "loss": 0.265, + "step": 3410 + }, + { + "epoch": 1.4912663755458515, + "grad_norm": 0.24516343045152553, + "learning_rate": 2.960365312818967e-05, + "loss": 0.2745, + "step": 3415 + }, + { + "epoch": 1.4934497816593886, + "grad_norm": 0.31298311253304867, + "learning_rate": 2.9541211768207883e-05, + "loss": 0.2718, + "step": 3420 + }, + { + "epoch": 1.4956331877729259, + "grad_norm": 0.3182623875469872, + "learning_rate": 2.9478741098431027e-05, + "loss": 0.2747, + "step": 3425 + }, + { + "epoch": 1.497816593886463, + "grad_norm": 0.31623459423577244, + "learning_rate": 2.9416241522055987e-05, + "loss": 0.3007, + "step": 3430 + }, + { + "epoch": 1.5, + "grad_norm": 0.3659591897091837, + "learning_rate": 2.9353713442466246e-05, + "loss": 0.2797, + "step": 3435 + }, + { + "epoch": 1.502183406113537, + "grad_norm": 0.29892370752426095, + "learning_rate": 2.929115726322924e-05, + "loss": 0.2711, + "step": 3440 + }, + { + "epoch": 1.5043668122270741, + "grad_norm": 0.36468217436567685, + "learning_rate": 2.9228573388093765e-05, + "loss": 0.2781, + "step": 3445 + }, + { + "epoch": 1.5065502183406112, + "grad_norm": 0.36523975525457747, + "learning_rate": 2.916596222098738e-05, + "loss": 0.2782, + "step": 3450 + }, + { + "epoch": 1.5087336244541485, + "grad_norm": 0.24799351938184713, + "learning_rate": 2.9103324166013777e-05, + "loss": 0.2723, + "step": 3455 + }, + { + "epoch": 1.5109170305676856, + "grad_norm": 0.3395645662027844, + "learning_rate": 2.9040659627450195e-05, + "loss": 0.2734, + "step": 3460 + }, + { + "epoch": 1.5131004366812228, + "grad_norm": 0.2312365794117943, + "learning_rate": 2.897796900974481e-05, + "loss": 0.2695, + "step": 3465 + }, + { + "epoch": 1.51528384279476, + "grad_norm": 0.2733485343314042, + "learning_rate": 2.8915252717514108e-05, + "loss": 0.2753, + "step": 3470 + }, + { + "epoch": 1.517467248908297, + "grad_norm": 0.30853988506735613, + "learning_rate": 2.8852511155540286e-05, + "loss": 0.2805, + "step": 3475 + }, + { + "epoch": 1.519650655021834, + "grad_norm": 0.298316972177523, + "learning_rate": 2.8789744728768632e-05, + "loss": 0.2674, + "step": 3480 + }, + { + "epoch": 1.5218340611353711, + "grad_norm": 0.33489826338103257, + "learning_rate": 2.8726953842304917e-05, + "loss": 0.2781, + "step": 3485 + }, + { + "epoch": 1.5240174672489082, + "grad_norm": 0.29180816733941406, + "learning_rate": 2.8664138901412784e-05, + "loss": 0.2828, + "step": 3490 + }, + { + "epoch": 1.5262008733624453, + "grad_norm": 0.2652928534046041, + "learning_rate": 2.8601300311511135e-05, + "loss": 0.2704, + "step": 3495 + }, + { + "epoch": 1.5283842794759825, + "grad_norm": 0.2478372696723595, + "learning_rate": 2.8538438478171493e-05, + "loss": 0.2709, + "step": 3500 + }, + { + "epoch": 1.5305676855895196, + "grad_norm": 0.29590356196390866, + "learning_rate": 2.8475553807115392e-05, + "loss": 0.2645, + "step": 3505 + }, + { + "epoch": 1.532751091703057, + "grad_norm": 0.2404106523342164, + "learning_rate": 2.8412646704211787e-05, + "loss": 0.2716, + "step": 3510 + }, + { + "epoch": 1.534934497816594, + "grad_norm": 0.2913911540734402, + "learning_rate": 2.8349717575474395e-05, + "loss": 0.2812, + "step": 3515 + }, + { + "epoch": 1.537117903930131, + "grad_norm": 0.27619563545275305, + "learning_rate": 2.8286766827059103e-05, + "loss": 0.2852, + "step": 3520 + }, + { + "epoch": 1.539301310043668, + "grad_norm": 0.25296146919214546, + "learning_rate": 2.8223794865261326e-05, + "loss": 0.2706, + "step": 3525 + }, + { + "epoch": 1.5414847161572052, + "grad_norm": 0.4120208095199771, + "learning_rate": 2.8160802096513396e-05, + "loss": 0.2655, + "step": 3530 + }, + { + "epoch": 1.5436681222707422, + "grad_norm": 0.2797662392218346, + "learning_rate": 2.809778892738194e-05, + "loss": 0.2799, + "step": 3535 + }, + { + "epoch": 1.5458515283842795, + "grad_norm": 0.30491897219029096, + "learning_rate": 2.8034755764565252e-05, + "loss": 0.2703, + "step": 3540 + }, + { + "epoch": 1.5480349344978166, + "grad_norm": 0.2830893746174212, + "learning_rate": 2.797170301489066e-05, + "loss": 0.2716, + "step": 3545 + }, + { + "epoch": 1.5502183406113537, + "grad_norm": 0.23485142267745251, + "learning_rate": 2.7908631085311933e-05, + "loss": 0.2617, + "step": 3550 + }, + { + "epoch": 1.552401746724891, + "grad_norm": 0.3021327029538882, + "learning_rate": 2.78455403829066e-05, + "loss": 0.2638, + "step": 3555 + }, + { + "epoch": 1.554585152838428, + "grad_norm": 0.2621866662981608, + "learning_rate": 2.7782431314873377e-05, + "loss": 0.2678, + "step": 3560 + }, + { + "epoch": 1.556768558951965, + "grad_norm": 0.2747099055332518, + "learning_rate": 2.7719304288529503e-05, + "loss": 0.2714, + "step": 3565 + }, + { + "epoch": 1.5589519650655022, + "grad_norm": 0.27602610129495325, + "learning_rate": 2.7656159711308127e-05, + "loss": 0.2754, + "step": 3570 + }, + { + "epoch": 1.5611353711790392, + "grad_norm": 0.2563630850581704, + "learning_rate": 2.7592997990755665e-05, + "loss": 0.2674, + "step": 3575 + }, + { + "epoch": 1.5633187772925763, + "grad_norm": 0.25276036483666414, + "learning_rate": 2.75298195345292e-05, + "loss": 0.2689, + "step": 3580 + }, + { + "epoch": 1.5655021834061136, + "grad_norm": 0.2568576656221897, + "learning_rate": 2.7466624750393804e-05, + "loss": 0.2624, + "step": 3585 + }, + { + "epoch": 1.5676855895196506, + "grad_norm": 0.2540380161679821, + "learning_rate": 2.740341404621996e-05, + "loss": 0.2704, + "step": 3590 + }, + { + "epoch": 1.569868995633188, + "grad_norm": 0.2691098927474828, + "learning_rate": 2.7340187829980884e-05, + "loss": 0.2596, + "step": 3595 + }, + { + "epoch": 1.572052401746725, + "grad_norm": 0.27708252166503045, + "learning_rate": 2.7276946509749905e-05, + "loss": 0.2753, + "step": 3600 + }, + { + "epoch": 1.574235807860262, + "grad_norm": 0.2828335584788292, + "learning_rate": 2.7213690493697852e-05, + "loss": 0.2813, + "step": 3605 + }, + { + "epoch": 1.5764192139737991, + "grad_norm": 0.29157807173338984, + "learning_rate": 2.7150420190090397e-05, + "loss": 0.2654, + "step": 3610 + }, + { + "epoch": 1.5786026200873362, + "grad_norm": 0.2621619308124079, + "learning_rate": 2.7087136007285417e-05, + "loss": 0.269, + "step": 3615 + }, + { + "epoch": 1.5807860262008733, + "grad_norm": 0.27055849925403364, + "learning_rate": 2.7023838353730386e-05, + "loss": 0.2603, + "step": 3620 + }, + { + "epoch": 1.5829694323144103, + "grad_norm": 0.28812667281668847, + "learning_rate": 2.6960527637959704e-05, + "loss": 0.2615, + "step": 3625 + }, + { + "epoch": 1.5851528384279476, + "grad_norm": 0.3648305241785186, + "learning_rate": 2.689720426859208e-05, + "loss": 0.2745, + "step": 3630 + }, + { + "epoch": 1.5873362445414847, + "grad_norm": 0.3579531717893759, + "learning_rate": 2.68338686543279e-05, + "loss": 0.2754, + "step": 3635 + }, + { + "epoch": 1.589519650655022, + "grad_norm": 0.30496840176534284, + "learning_rate": 2.6770521203946582e-05, + "loss": 0.2687, + "step": 3640 + }, + { + "epoch": 1.591703056768559, + "grad_norm": 0.3316407097346181, + "learning_rate": 2.670716232630392e-05, + "loss": 0.2708, + "step": 3645 + }, + { + "epoch": 1.5938864628820961, + "grad_norm": 0.30884262120013556, + "learning_rate": 2.6643792430329473e-05, + "loss": 0.2691, + "step": 3650 + }, + { + "epoch": 1.5960698689956332, + "grad_norm": 0.24575164955804643, + "learning_rate": 2.6580411925023916e-05, + "loss": 0.2895, + "step": 3655 + }, + { + "epoch": 1.5982532751091703, + "grad_norm": 0.2403162634268798, + "learning_rate": 2.6517021219456402e-05, + "loss": 0.2768, + "step": 3660 + }, + { + "epoch": 1.6004366812227073, + "grad_norm": 0.7286289540628894, + "learning_rate": 2.6453620722761896e-05, + "loss": 0.2772, + "step": 3665 + }, + { + "epoch": 1.6026200873362444, + "grad_norm": 0.2698469031360812, + "learning_rate": 2.6390210844138593e-05, + "loss": 0.2772, + "step": 3670 + }, + { + "epoch": 1.6048034934497817, + "grad_norm": 0.29594772177235845, + "learning_rate": 2.6326791992845202e-05, + "loss": 0.2746, + "step": 3675 + }, + { + "epoch": 1.6069868995633187, + "grad_norm": 0.3432286485907405, + "learning_rate": 2.626336457819837e-05, + "loss": 0.2627, + "step": 3680 + }, + { + "epoch": 1.609170305676856, + "grad_norm": 0.47848404072485456, + "learning_rate": 2.619992900957e-05, + "loss": 0.2896, + "step": 3685 + }, + { + "epoch": 1.611353711790393, + "grad_norm": 0.400345398790263, + "learning_rate": 2.613648569638464e-05, + "loss": 0.2832, + "step": 3690 + }, + { + "epoch": 1.6135371179039302, + "grad_norm": 0.26168308925039074, + "learning_rate": 2.6073035048116796e-05, + "loss": 0.2693, + "step": 3695 + }, + { + "epoch": 1.6157205240174672, + "grad_norm": 0.33192980592740756, + "learning_rate": 2.6009577474288333e-05, + "loss": 0.2759, + "step": 3700 + }, + { + "epoch": 1.6179039301310043, + "grad_norm": 0.2787526637545283, + "learning_rate": 2.5946113384465813e-05, + "loss": 0.2642, + "step": 3705 + }, + { + "epoch": 1.6200873362445414, + "grad_norm": 0.27338870781942926, + "learning_rate": 2.5882643188257856e-05, + "loss": 0.2706, + "step": 3710 + }, + { + "epoch": 1.6222707423580784, + "grad_norm": 0.2435797974873236, + "learning_rate": 2.5819167295312485e-05, + "loss": 0.2724, + "step": 3715 + }, + { + "epoch": 1.6244541484716157, + "grad_norm": 0.29781182108189724, + "learning_rate": 2.57556861153145e-05, + "loss": 0.2707, + "step": 3720 + }, + { + "epoch": 1.6266375545851528, + "grad_norm": 0.32478484905425753, + "learning_rate": 2.569220005798281e-05, + "loss": 0.2725, + "step": 3725 + }, + { + "epoch": 1.62882096069869, + "grad_norm": 0.2669336390561608, + "learning_rate": 2.5628709533067824e-05, + "loss": 0.2713, + "step": 3730 + }, + { + "epoch": 1.6310043668122272, + "grad_norm": 0.2573882639282421, + "learning_rate": 2.556521495034877e-05, + "loss": 0.2606, + "step": 3735 + }, + { + "epoch": 1.6331877729257642, + "grad_norm": 0.2705880029160448, + "learning_rate": 2.550171671963108e-05, + "loss": 0.2575, + "step": 3740 + }, + { + "epoch": 1.6353711790393013, + "grad_norm": 0.27796104371371233, + "learning_rate": 2.543821525074371e-05, + "loss": 0.2699, + "step": 3745 + }, + { + "epoch": 1.6375545851528384, + "grad_norm": 0.2968772131918911, + "learning_rate": 2.537471095353653e-05, + "loss": 0.2604, + "step": 3750 + }, + { + "epoch": 1.6397379912663754, + "grad_norm": 0.30473359891717344, + "learning_rate": 2.5311204237877657e-05, + "loss": 0.2715, + "step": 3755 + }, + { + "epoch": 1.6419213973799127, + "grad_norm": 0.29986937512458306, + "learning_rate": 2.524769551365083e-05, + "loss": 0.2569, + "step": 3760 + }, + { + "epoch": 1.6441048034934498, + "grad_norm": 0.29756570977438257, + "learning_rate": 2.5184185190752752e-05, + "loss": 0.2805, + "step": 3765 + }, + { + "epoch": 1.6462882096069869, + "grad_norm": 0.24764879064376663, + "learning_rate": 2.512067367909041e-05, + "loss": 0.2723, + "step": 3770 + }, + { + "epoch": 1.6484716157205241, + "grad_norm": 0.2536796536666786, + "learning_rate": 2.505716138857851e-05, + "loss": 0.2692, + "step": 3775 + }, + { + "epoch": 1.6506550218340612, + "grad_norm": 0.27985489044371364, + "learning_rate": 2.499364872913676e-05, + "loss": 0.2736, + "step": 3780 + }, + { + "epoch": 1.6528384279475983, + "grad_norm": 0.24159737014671576, + "learning_rate": 2.4930136110687246e-05, + "loss": 0.2711, + "step": 3785 + }, + { + "epoch": 1.6550218340611353, + "grad_norm": 0.28234624412572884, + "learning_rate": 2.486662394315181e-05, + "loss": 0.2865, + "step": 3790 + }, + { + "epoch": 1.6572052401746724, + "grad_norm": 0.2491348077927823, + "learning_rate": 2.4803112636449355e-05, + "loss": 0.2731, + "step": 3795 + }, + { + "epoch": 1.6593886462882095, + "grad_norm": 0.31780593847271627, + "learning_rate": 2.473960260049325e-05, + "loss": 0.2789, + "step": 3800 + }, + { + "epoch": 1.6615720524017468, + "grad_norm": 0.2685872565371744, + "learning_rate": 2.4676094245188658e-05, + "loss": 0.2933, + "step": 3805 + }, + { + "epoch": 1.6637554585152838, + "grad_norm": 0.263863501194976, + "learning_rate": 2.461258798042989e-05, + "loss": 0.2685, + "step": 3810 + }, + { + "epoch": 1.6659388646288211, + "grad_norm": 0.21766584856793741, + "learning_rate": 2.454908421609777e-05, + "loss": 0.2552, + "step": 3815 + }, + { + "epoch": 1.6681222707423582, + "grad_norm": 0.27428707449471423, + "learning_rate": 2.4485583362056974e-05, + "loss": 0.2786, + "step": 3820 + }, + { + "epoch": 1.6703056768558953, + "grad_norm": 0.2835969591758734, + "learning_rate": 2.4422085828153403e-05, + "loss": 0.2662, + "step": 3825 + }, + { + "epoch": 1.6724890829694323, + "grad_norm": 0.29526168010306597, + "learning_rate": 2.435859202421153e-05, + "loss": 0.269, + "step": 3830 + }, + { + "epoch": 1.6746724890829694, + "grad_norm": 0.30337514206743876, + "learning_rate": 2.429510236003175e-05, + "loss": 0.2597, + "step": 3835 + }, + { + "epoch": 1.6768558951965065, + "grad_norm": 0.2584878592869018, + "learning_rate": 2.4231617245387745e-05, + "loss": 0.2709, + "step": 3840 + }, + { + "epoch": 1.6790393013100435, + "grad_norm": 0.27852560485634303, + "learning_rate": 2.416813709002382e-05, + "loss": 0.2712, + "step": 3845 + }, + { + "epoch": 1.6812227074235808, + "grad_norm": 0.31941354537015787, + "learning_rate": 2.4104662303652288e-05, + "loss": 0.286, + "step": 3850 + }, + { + "epoch": 1.6834061135371179, + "grad_norm": 0.2639262509863191, + "learning_rate": 2.4041193295950788e-05, + "loss": 0.2679, + "step": 3855 + }, + { + "epoch": 1.6855895196506552, + "grad_norm": 0.27648769301544285, + "learning_rate": 2.39777304765597e-05, + "loss": 0.2764, + "step": 3860 + }, + { + "epoch": 1.6877729257641922, + "grad_norm": 0.2557231698124579, + "learning_rate": 2.391427425507943e-05, + "loss": 0.2773, + "step": 3865 + }, + { + "epoch": 1.6899563318777293, + "grad_norm": 0.2835965272930869, + "learning_rate": 2.385082504106781e-05, + "loss": 0.2655, + "step": 3870 + }, + { + "epoch": 1.6921397379912664, + "grad_norm": 0.30219688777437337, + "learning_rate": 2.378738324403744e-05, + "loss": 0.2649, + "step": 3875 + }, + { + "epoch": 1.6943231441048034, + "grad_norm": 0.286618784377373, + "learning_rate": 2.3723949273453073e-05, + "loss": 0.2668, + "step": 3880 + }, + { + "epoch": 1.6965065502183405, + "grad_norm": 0.31422639752909204, + "learning_rate": 2.3660523538728904e-05, + "loss": 0.2671, + "step": 3885 + }, + { + "epoch": 1.6986899563318776, + "grad_norm": 0.24899445840743908, + "learning_rate": 2.3597106449226015e-05, + "loss": 0.2603, + "step": 3890 + }, + { + "epoch": 1.7008733624454149, + "grad_norm": 0.28778299708675603, + "learning_rate": 2.3533698414249668e-05, + "loss": 0.2532, + "step": 3895 + }, + { + "epoch": 1.703056768558952, + "grad_norm": 0.3010236993612902, + "learning_rate": 2.3470299843046693e-05, + "loss": 0.2607, + "step": 3900 + }, + { + "epoch": 1.7052401746724892, + "grad_norm": 0.2505565783956947, + "learning_rate": 2.340691114480284e-05, + "loss": 0.2675, + "step": 3905 + }, + { + "epoch": 1.7074235807860263, + "grad_norm": 0.2534199791327587, + "learning_rate": 2.334353272864013e-05, + "loss": 0.2727, + "step": 3910 + }, + { + "epoch": 1.7096069868995634, + "grad_norm": 0.2759471253942165, + "learning_rate": 2.3280165003614227e-05, + "loss": 0.2663, + "step": 3915 + }, + { + "epoch": 1.7117903930131004, + "grad_norm": 0.2958147406172001, + "learning_rate": 2.3216808378711788e-05, + "loss": 0.2705, + "step": 3920 + }, + { + "epoch": 1.7139737991266375, + "grad_norm": 0.26156474466311214, + "learning_rate": 2.3153463262847835e-05, + "loss": 0.242, + "step": 3925 + }, + { + "epoch": 1.7161572052401746, + "grad_norm": 0.29968058749209264, + "learning_rate": 2.3090130064863113e-05, + "loss": 0.2681, + "step": 3930 + }, + { + "epoch": 1.7183406113537116, + "grad_norm": 0.24513602810052307, + "learning_rate": 2.3026809193521423e-05, + "loss": 0.2746, + "step": 3935 + }, + { + "epoch": 1.720524017467249, + "grad_norm": 0.2624738237172384, + "learning_rate": 2.296350105750703e-05, + "loss": 0.2682, + "step": 3940 + }, + { + "epoch": 1.722707423580786, + "grad_norm": 0.21915397316253904, + "learning_rate": 2.2900206065421997e-05, + "loss": 0.26, + "step": 3945 + }, + { + "epoch": 1.7248908296943233, + "grad_norm": 0.248859562817194, + "learning_rate": 2.2836924625783554e-05, + "loss": 0.2776, + "step": 3950 + }, + { + "epoch": 1.7270742358078603, + "grad_norm": 0.2941037210948601, + "learning_rate": 2.2773657147021464e-05, + "loss": 0.281, + "step": 3955 + }, + { + "epoch": 1.7292576419213974, + "grad_norm": 0.22756003553556517, + "learning_rate": 2.271040403747537e-05, + "loss": 0.2597, + "step": 3960 + }, + { + "epoch": 1.7314410480349345, + "grad_norm": 0.274949859317834, + "learning_rate": 2.2647165705392186e-05, + "loss": 0.2697, + "step": 3965 + }, + { + "epoch": 1.7336244541484715, + "grad_norm": 0.2797174304909422, + "learning_rate": 2.2583942558923456e-05, + "loss": 0.2716, + "step": 3970 + }, + { + "epoch": 1.7358078602620086, + "grad_norm": 0.26704187553881054, + "learning_rate": 2.2520735006122688e-05, + "loss": 0.274, + "step": 3975 + }, + { + "epoch": 1.737991266375546, + "grad_norm": 0.284507084979283, + "learning_rate": 2.2457543454942786e-05, + "loss": 0.2587, + "step": 3980 + }, + { + "epoch": 1.740174672489083, + "grad_norm": 0.2699117094732979, + "learning_rate": 2.2394368313233332e-05, + "loss": 0.2619, + "step": 3985 + }, + { + "epoch": 1.74235807860262, + "grad_norm": 0.25250488525353865, + "learning_rate": 2.2331209988738024e-05, + "loss": 0.2602, + "step": 3990 + }, + { + "epoch": 1.7445414847161573, + "grad_norm": 0.24206728711747397, + "learning_rate": 2.226806888909202e-05, + "loss": 0.262, + "step": 3995 + }, + { + "epoch": 1.7467248908296944, + "grad_norm": 0.2907832129799141, + "learning_rate": 2.2204945421819292e-05, + "loss": 0.2636, + "step": 4000 + }, + { + "epoch": 1.7489082969432315, + "grad_norm": 0.29699912162411024, + "learning_rate": 2.2141839994330027e-05, + "loss": 0.2747, + "step": 4005 + }, + { + "epoch": 1.7510917030567685, + "grad_norm": 0.2505983156960274, + "learning_rate": 2.207875301391796e-05, + "loss": 0.2706, + "step": 4010 + }, + { + "epoch": 1.7532751091703056, + "grad_norm": 0.20887238006814582, + "learning_rate": 2.2015684887757776e-05, + "loss": 0.2544, + "step": 4015 + }, + { + "epoch": 1.7554585152838427, + "grad_norm": 0.2350828605757233, + "learning_rate": 2.195263602290248e-05, + "loss": 0.2759, + "step": 4020 + }, + { + "epoch": 1.75764192139738, + "grad_norm": 0.27919466794671643, + "learning_rate": 2.188960682628074e-05, + "loss": 0.2468, + "step": 4025 + }, + { + "epoch": 1.759825327510917, + "grad_norm": 0.24755062875053999, + "learning_rate": 2.1826597704694306e-05, + "loss": 0.2699, + "step": 4030 + }, + { + "epoch": 1.762008733624454, + "grad_norm": 0.28930040508547633, + "learning_rate": 2.176360906481533e-05, + "loss": 0.27, + "step": 4035 + }, + { + "epoch": 1.7641921397379914, + "grad_norm": 0.2463760850445127, + "learning_rate": 2.1700641313183807e-05, + "loss": 0.2604, + "step": 4040 + }, + { + "epoch": 1.7663755458515285, + "grad_norm": 0.24878427896848646, + "learning_rate": 2.163769485620489e-05, + "loss": 0.2647, + "step": 4045 + }, + { + "epoch": 1.7685589519650655, + "grad_norm": 0.3491349213382724, + "learning_rate": 2.1574770100146295e-05, + "loss": 0.2717, + "step": 4050 + }, + { + "epoch": 1.7707423580786026, + "grad_norm": 0.26013086945480984, + "learning_rate": 2.1511867451135693e-05, + "loss": 0.2599, + "step": 4055 + }, + { + "epoch": 1.7729257641921397, + "grad_norm": 0.2900875329162927, + "learning_rate": 2.1448987315158048e-05, + "loss": 0.2686, + "step": 4060 + }, + { + "epoch": 1.7751091703056767, + "grad_norm": 0.2653600164327816, + "learning_rate": 2.1386130098053036e-05, + "loss": 0.2627, + "step": 4065 + }, + { + "epoch": 1.777292576419214, + "grad_norm": 0.27303311100886735, + "learning_rate": 2.1323296205512418e-05, + "loss": 0.2708, + "step": 4070 + }, + { + "epoch": 1.779475982532751, + "grad_norm": 0.29210046654985766, + "learning_rate": 2.1260486043077388e-05, + "loss": 0.2581, + "step": 4075 + }, + { + "epoch": 1.7816593886462884, + "grad_norm": 0.284149374163666, + "learning_rate": 2.1197700016136014e-05, + "loss": 0.266, + "step": 4080 + }, + { + "epoch": 1.7838427947598254, + "grad_norm": 0.2409781119667348, + "learning_rate": 2.113493852992056e-05, + "loss": 0.2571, + "step": 4085 + }, + { + "epoch": 1.7860262008733625, + "grad_norm": 0.20843436142983807, + "learning_rate": 2.1072201989504913e-05, + "loss": 0.2733, + "step": 4090 + }, + { + "epoch": 1.7882096069868996, + "grad_norm": 0.23230387555929172, + "learning_rate": 2.1009490799801965e-05, + "loss": 0.269, + "step": 4095 + }, + { + "epoch": 1.7903930131004366, + "grad_norm": 0.2694983772096665, + "learning_rate": 2.0946805365560966e-05, + "loss": 0.2745, + "step": 4100 + }, + { + "epoch": 1.7925764192139737, + "grad_norm": 0.24747448881142187, + "learning_rate": 2.0884146091364956e-05, + "loss": 0.2659, + "step": 4105 + }, + { + "epoch": 1.7947598253275108, + "grad_norm": 0.22323270338570358, + "learning_rate": 2.0821513381628132e-05, + "loss": 0.2645, + "step": 4110 + }, + { + "epoch": 1.796943231441048, + "grad_norm": 0.2580592437636737, + "learning_rate": 2.0758907640593228e-05, + "loss": 0.2563, + "step": 4115 + }, + { + "epoch": 1.7991266375545851, + "grad_norm": 0.2411641382308118, + "learning_rate": 2.0696329272328932e-05, + "loss": 0.2721, + "step": 4120 + }, + { + "epoch": 1.8013100436681224, + "grad_norm": 0.25806558253779854, + "learning_rate": 2.063377868072724e-05, + "loss": 0.2616, + "step": 4125 + }, + { + "epoch": 1.8034934497816595, + "grad_norm": 0.2455964016028092, + "learning_rate": 2.0571256269500893e-05, + "loss": 0.271, + "step": 4130 + }, + { + "epoch": 1.8056768558951966, + "grad_norm": 0.25469457274681495, + "learning_rate": 2.0508762442180745e-05, + "loss": 0.2683, + "step": 4135 + }, + { + "epoch": 1.8078602620087336, + "grad_norm": 0.2734524817312364, + "learning_rate": 2.0446297602113167e-05, + "loss": 0.2782, + "step": 4140 + }, + { + "epoch": 1.8100436681222707, + "grad_norm": 0.2711480537954455, + "learning_rate": 2.0383862152457432e-05, + "loss": 0.2636, + "step": 4145 + }, + { + "epoch": 1.8122270742358078, + "grad_norm": 0.23192737600539587, + "learning_rate": 2.0321456496183134e-05, + "loss": 0.276, + "step": 4150 + }, + { + "epoch": 1.8144104803493448, + "grad_norm": 0.2610658805286943, + "learning_rate": 2.0259081036067563e-05, + "loss": 0.2706, + "step": 4155 + }, + { + "epoch": 1.8165938864628821, + "grad_norm": 0.2670079724190112, + "learning_rate": 2.0196736174693126e-05, + "loss": 0.2529, + "step": 4160 + }, + { + "epoch": 1.8187772925764192, + "grad_norm": 0.25454455186632396, + "learning_rate": 2.013442231444474e-05, + "loss": 0.2649, + "step": 4165 + }, + { + "epoch": 1.8209606986899565, + "grad_norm": 0.2777703715553109, + "learning_rate": 2.007213985750724e-05, + "loss": 0.265, + "step": 4170 + }, + { + "epoch": 1.8231441048034935, + "grad_norm": 0.2654974710451674, + "learning_rate": 2.000988920586275e-05, + "loss": 0.2653, + "step": 4175 + }, + { + "epoch": 1.8253275109170306, + "grad_norm": 0.28984168869856813, + "learning_rate": 1.994767076128816e-05, + "loss": 0.2572, + "step": 4180 + }, + { + "epoch": 1.8275109170305677, + "grad_norm": 0.2620326885283614, + "learning_rate": 1.988548492535246e-05, + "loss": 0.2754, + "step": 4185 + }, + { + "epoch": 1.8296943231441047, + "grad_norm": 0.24235633152361688, + "learning_rate": 1.9823332099414185e-05, + "loss": 0.2717, + "step": 4190 + }, + { + "epoch": 1.8318777292576418, + "grad_norm": 0.24991695967334396, + "learning_rate": 1.9761212684618824e-05, + "loss": 0.25, + "step": 4195 + }, + { + "epoch": 1.8340611353711789, + "grad_norm": 0.21335483219532675, + "learning_rate": 1.9699127081896214e-05, + "loss": 0.2513, + "step": 4200 + }, + { + "epoch": 1.8362445414847162, + "grad_norm": 0.2515339708285586, + "learning_rate": 1.963707569195797e-05, + "loss": 0.2524, + "step": 4205 + }, + { + "epoch": 1.8384279475982532, + "grad_norm": 0.23802145795043225, + "learning_rate": 1.9575058915294885e-05, + "loss": 0.2652, + "step": 4210 + }, + { + "epoch": 1.8406113537117905, + "grad_norm": 0.30056095101017133, + "learning_rate": 1.9513077152174363e-05, + "loss": 0.2776, + "step": 4215 + }, + { + "epoch": 1.8427947598253276, + "grad_norm": 0.2915453674361558, + "learning_rate": 1.9451130802637822e-05, + "loss": 0.2588, + "step": 4220 + }, + { + "epoch": 1.8449781659388647, + "grad_norm": 0.30777863799002625, + "learning_rate": 1.9389220266498102e-05, + "loss": 0.2701, + "step": 4225 + }, + { + "epoch": 1.8471615720524017, + "grad_norm": 0.2597777533142635, + "learning_rate": 1.9327345943336904e-05, + "loss": 0.2645, + "step": 4230 + }, + { + "epoch": 1.8493449781659388, + "grad_norm": 0.3080444941319507, + "learning_rate": 1.9265508232502208e-05, + "loss": 0.2749, + "step": 4235 + }, + { + "epoch": 1.8515283842794759, + "grad_norm": 0.2831226100931665, + "learning_rate": 1.9203707533105687e-05, + "loss": 0.2598, + "step": 4240 + }, + { + "epoch": 1.8537117903930131, + "grad_norm": 0.27124968567558605, + "learning_rate": 1.914194424402015e-05, + "loss": 0.271, + "step": 4245 + }, + { + "epoch": 1.8558951965065502, + "grad_norm": 0.24173431823915756, + "learning_rate": 1.9080218763876917e-05, + "loss": 0.2512, + "step": 4250 + }, + { + "epoch": 1.8580786026200873, + "grad_norm": 0.27847599015903607, + "learning_rate": 1.901853149106332e-05, + "loss": 0.263, + "step": 4255 + }, + { + "epoch": 1.8602620087336246, + "grad_norm": 0.348403165189276, + "learning_rate": 1.8956882823720075e-05, + "loss": 0.2657, + "step": 4260 + }, + { + "epoch": 1.8624454148471616, + "grad_norm": 0.24233816768582403, + "learning_rate": 1.889527315973874e-05, + "loss": 0.26, + "step": 4265 + }, + { + "epoch": 1.8646288209606987, + "grad_norm": 0.23840655638016398, + "learning_rate": 1.8833702896759132e-05, + "loss": 0.2636, + "step": 4270 + }, + { + "epoch": 1.8668122270742358, + "grad_norm": 0.2549222906252635, + "learning_rate": 1.8772172432166767e-05, + "loss": 0.2744, + "step": 4275 + }, + { + "epoch": 1.8689956331877728, + "grad_norm": 0.25475120516780736, + "learning_rate": 1.87106821630903e-05, + "loss": 0.2748, + "step": 4280 + }, + { + "epoch": 1.87117903930131, + "grad_norm": 0.24088260430162353, + "learning_rate": 1.8649232486398955e-05, + "loss": 0.2638, + "step": 4285 + }, + { + "epoch": 1.8733624454148472, + "grad_norm": 0.2556026511315039, + "learning_rate": 1.8587823798699958e-05, + "loss": 0.2601, + "step": 4290 + }, + { + "epoch": 1.8755458515283843, + "grad_norm": 0.2700718784491458, + "learning_rate": 1.8526456496335997e-05, + "loss": 0.2525, + "step": 4295 + }, + { + "epoch": 1.8777292576419216, + "grad_norm": 0.27667715242466545, + "learning_rate": 1.846513097538265e-05, + "loss": 0.2547, + "step": 4300 + }, + { + "epoch": 1.8799126637554586, + "grad_norm": 0.2488594588357838, + "learning_rate": 1.8403847631645815e-05, + "loss": 0.2747, + "step": 4305 + }, + { + "epoch": 1.8820960698689957, + "grad_norm": 0.2486103000672603, + "learning_rate": 1.834260686065919e-05, + "loss": 0.2755, + "step": 4310 + }, + { + "epoch": 1.8842794759825328, + "grad_norm": 0.2673003900981902, + "learning_rate": 1.8281409057681685e-05, + "loss": 0.2556, + "step": 4315 + }, + { + "epoch": 1.8864628820960698, + "grad_norm": 0.27816423985109073, + "learning_rate": 1.8220254617694904e-05, + "loss": 0.2725, + "step": 4320 + }, + { + "epoch": 1.888646288209607, + "grad_norm": 0.21891089387179588, + "learning_rate": 1.815914393540057e-05, + "loss": 0.2672, + "step": 4325 + }, + { + "epoch": 1.890829694323144, + "grad_norm": 0.3027560989030244, + "learning_rate": 1.8098077405217973e-05, + "loss": 0.2681, + "step": 4330 + }, + { + "epoch": 1.8930131004366813, + "grad_norm": 0.2956885379322635, + "learning_rate": 1.8037055421281453e-05, + "loss": 0.2754, + "step": 4335 + }, + { + "epoch": 1.8951965065502183, + "grad_norm": 0.27344213459308064, + "learning_rate": 1.7976078377437843e-05, + "loss": 0.254, + "step": 4340 + }, + { + "epoch": 1.8973799126637556, + "grad_norm": 0.25908804251433426, + "learning_rate": 1.7915146667243912e-05, + "loss": 0.2709, + "step": 4345 + }, + { + "epoch": 1.8995633187772927, + "grad_norm": 0.27387774174550766, + "learning_rate": 1.785426068396385e-05, + "loss": 0.2724, + "step": 4350 + }, + { + "epoch": 1.9017467248908297, + "grad_norm": 0.2349626027532333, + "learning_rate": 1.7793420820566698e-05, + "loss": 0.2685, + "step": 4355 + }, + { + "epoch": 1.9039301310043668, + "grad_norm": 0.27519209684053136, + "learning_rate": 1.7732627469723867e-05, + "loss": 0.2665, + "step": 4360 + }, + { + "epoch": 1.9061135371179039, + "grad_norm": 0.2237579472013528, + "learning_rate": 1.7671881023806526e-05, + "loss": 0.2629, + "step": 4365 + }, + { + "epoch": 1.908296943231441, + "grad_norm": 0.24463587078424295, + "learning_rate": 1.7611181874883133e-05, + "loss": 0.2637, + "step": 4370 + }, + { + "epoch": 1.910480349344978, + "grad_norm": 0.30256331204318965, + "learning_rate": 1.755053041471689e-05, + "loss": 0.2626, + "step": 4375 + }, + { + "epoch": 1.9126637554585153, + "grad_norm": 0.23702938555275288, + "learning_rate": 1.748992703476318e-05, + "loss": 0.2621, + "step": 4380 + }, + { + "epoch": 1.9148471615720524, + "grad_norm": 0.29244916287768213, + "learning_rate": 1.7429372126167105e-05, + "loss": 0.2492, + "step": 4385 + }, + { + "epoch": 1.9170305676855897, + "grad_norm": 0.2588931757550786, + "learning_rate": 1.7368866079760887e-05, + "loss": 0.2615, + "step": 4390 + }, + { + "epoch": 1.9192139737991267, + "grad_norm": 0.25196463301087313, + "learning_rate": 1.7308409286061407e-05, + "loss": 0.2686, + "step": 4395 + }, + { + "epoch": 1.9213973799126638, + "grad_norm": 0.2158443773081268, + "learning_rate": 1.7248002135267645e-05, + "loss": 0.2666, + "step": 4400 + }, + { + "epoch": 1.9235807860262009, + "grad_norm": 0.287220788904416, + "learning_rate": 1.7187645017258196e-05, + "loss": 0.2735, + "step": 4405 + }, + { + "epoch": 1.925764192139738, + "grad_norm": 0.23107125712315182, + "learning_rate": 1.712733832158872e-05, + "loss": 0.2613, + "step": 4410 + }, + { + "epoch": 1.927947598253275, + "grad_norm": 0.24853969589485442, + "learning_rate": 1.7067082437489436e-05, + "loss": 0.2653, + "step": 4415 + }, + { + "epoch": 1.930131004366812, + "grad_norm": 0.2767451923179264, + "learning_rate": 1.7006877753862632e-05, + "loss": 0.262, + "step": 4420 + }, + { + "epoch": 1.9323144104803494, + "grad_norm": 0.22927033548053927, + "learning_rate": 1.6946724659280133e-05, + "loss": 0.2542, + "step": 4425 + }, + { + "epoch": 1.9344978165938864, + "grad_norm": 0.23110650040140782, + "learning_rate": 1.6886623541980796e-05, + "loss": 0.2757, + "step": 4430 + }, + { + "epoch": 1.9366812227074237, + "grad_norm": 0.23731032145161848, + "learning_rate": 1.682657478986801e-05, + "loss": 0.2645, + "step": 4435 + }, + { + "epoch": 1.9388646288209608, + "grad_norm": 0.25552976383740716, + "learning_rate": 1.6766578790507192e-05, + "loss": 0.2595, + "step": 4440 + }, + { + "epoch": 1.9410480349344978, + "grad_norm": 0.2155945620763327, + "learning_rate": 1.6706635931123277e-05, + "loss": 0.2596, + "step": 4445 + }, + { + "epoch": 1.943231441048035, + "grad_norm": 0.24247256828652752, + "learning_rate": 1.664674659859822e-05, + "loss": 0.2493, + "step": 4450 + }, + { + "epoch": 1.945414847161572, + "grad_norm": 0.2516593577449982, + "learning_rate": 1.658691117946853e-05, + "loss": 0.2761, + "step": 4455 + }, + { + "epoch": 1.947598253275109, + "grad_norm": 0.21991257872575132, + "learning_rate": 1.6527130059922717e-05, + "loss": 0.2632, + "step": 4460 + }, + { + "epoch": 1.9497816593886463, + "grad_norm": 0.261259312013636, + "learning_rate": 1.646740362579886e-05, + "loss": 0.2615, + "step": 4465 + }, + { + "epoch": 1.9519650655021834, + "grad_norm": 0.2318715733570351, + "learning_rate": 1.6407732262582067e-05, + "loss": 0.2654, + "step": 4470 + }, + { + "epoch": 1.9541484716157205, + "grad_norm": 0.22989633286543176, + "learning_rate": 1.634811635540201e-05, + "loss": 0.2593, + "step": 4475 + }, + { + "epoch": 1.9563318777292578, + "grad_norm": 0.24017270832452148, + "learning_rate": 1.6288556289030452e-05, + "loss": 0.2676, + "step": 4480 + }, + { + "epoch": 1.9585152838427948, + "grad_norm": 0.25142686822052845, + "learning_rate": 1.6229052447878743e-05, + "loss": 0.2541, + "step": 4485 + }, + { + "epoch": 1.960698689956332, + "grad_norm": 0.22732754650526263, + "learning_rate": 1.6169605215995345e-05, + "loss": 0.259, + "step": 4490 + }, + { + "epoch": 1.962882096069869, + "grad_norm": 0.24853011047008652, + "learning_rate": 1.6110214977063345e-05, + "loss": 0.2559, + "step": 4495 + }, + { + "epoch": 1.965065502183406, + "grad_norm": 0.2557640948437553, + "learning_rate": 1.6050882114398e-05, + "loss": 0.2508, + "step": 4500 + }, + { + "epoch": 1.967248908296943, + "grad_norm": 0.24324444806418238, + "learning_rate": 1.5991607010944242e-05, + "loss": 0.2607, + "step": 4505 + }, + { + "epoch": 1.9694323144104804, + "grad_norm": 0.23773093829535516, + "learning_rate": 1.5932390049274227e-05, + "loss": 0.2644, + "step": 4510 + }, + { + "epoch": 1.9716157205240175, + "grad_norm": 0.2532308876036313, + "learning_rate": 1.587323161158484e-05, + "loss": 0.272, + "step": 4515 + }, + { + "epoch": 1.9737991266375547, + "grad_norm": 0.2617214304004118, + "learning_rate": 1.5814132079695232e-05, + "loss": 0.2649, + "step": 4520 + }, + { + "epoch": 1.9759825327510918, + "grad_norm": 0.27129287136393265, + "learning_rate": 1.57550918350444e-05, + "loss": 0.2591, + "step": 4525 + }, + { + "epoch": 1.9781659388646289, + "grad_norm": 0.24468851478815726, + "learning_rate": 1.5696111258688657e-05, + "loss": 0.256, + "step": 4530 + }, + { + "epoch": 1.980349344978166, + "grad_norm": 0.23891980205993738, + "learning_rate": 1.563719073129923e-05, + "loss": 0.2655, + "step": 4535 + }, + { + "epoch": 1.982532751091703, + "grad_norm": 0.21564016827597185, + "learning_rate": 1.557833063315976e-05, + "loss": 0.2593, + "step": 4540 + }, + { + "epoch": 1.98471615720524, + "grad_norm": 0.2744337627938704, + "learning_rate": 1.5519531344163873e-05, + "loss": 0.2593, + "step": 4545 + }, + { + "epoch": 1.9868995633187772, + "grad_norm": 0.239539464353039, + "learning_rate": 1.5460793243812732e-05, + "loss": 0.2645, + "step": 4550 + }, + { + "epoch": 1.9890829694323144, + "grad_norm": 0.2652718460025595, + "learning_rate": 1.5402116711212566e-05, + "loss": 0.2712, + "step": 4555 + }, + { + "epoch": 1.9912663755458515, + "grad_norm": 0.23601629641684965, + "learning_rate": 1.5343502125072236e-05, + "loss": 0.2542, + "step": 4560 + }, + { + "epoch": 1.9934497816593888, + "grad_norm": 0.23315305328846694, + "learning_rate": 1.5284949863700805e-05, + "loss": 0.2663, + "step": 4565 + }, + { + "epoch": 1.9956331877729259, + "grad_norm": 0.23782090171706022, + "learning_rate": 1.5226460305005044e-05, + "loss": 0.2622, + "step": 4570 + }, + { + "epoch": 1.997816593886463, + "grad_norm": 0.22388379965681177, + "learning_rate": 1.5168033826487072e-05, + "loss": 0.2692, + "step": 4575 + }, + { + "epoch": 2.0, + "grad_norm": 0.27446739099024114, + "learning_rate": 1.5109670805241846e-05, + "loss": 0.2572, + "step": 4580 + }, + { + "epoch": 2.002183406113537, + "grad_norm": 0.29167629982584226, + "learning_rate": 1.505137161795478e-05, + "loss": 0.2211, + "step": 4585 + }, + { + "epoch": 2.004366812227074, + "grad_norm": 0.2877855784905206, + "learning_rate": 1.4993136640899286e-05, + "loss": 0.2147, + "step": 4590 + }, + { + "epoch": 2.006550218340611, + "grad_norm": 0.2589731295275126, + "learning_rate": 1.4934966249934334e-05, + "loss": 0.2052, + "step": 4595 + }, + { + "epoch": 2.0087336244541483, + "grad_norm": 0.24025382992800187, + "learning_rate": 1.487686082050207e-05, + "loss": 0.2162, + "step": 4600 + }, + { + "epoch": 2.010917030567686, + "grad_norm": 0.22612669576702268, + "learning_rate": 1.4818820727625348e-05, + "loss": 0.2071, + "step": 4605 + }, + { + "epoch": 2.013100436681223, + "grad_norm": 0.2212365851012379, + "learning_rate": 1.4760846345905344e-05, + "loss": 0.2139, + "step": 4610 + }, + { + "epoch": 2.01528384279476, + "grad_norm": 0.20584034227093886, + "learning_rate": 1.4702938049519105e-05, + "loss": 0.2168, + "step": 4615 + }, + { + "epoch": 2.017467248908297, + "grad_norm": 0.21936675852848891, + "learning_rate": 1.4645096212217167e-05, + "loss": 0.2073, + "step": 4620 + }, + { + "epoch": 2.019650655021834, + "grad_norm": 0.26100976524835695, + "learning_rate": 1.4587321207321115e-05, + "loss": 0.2233, + "step": 4625 + }, + { + "epoch": 2.021834061135371, + "grad_norm": 0.2056350483707369, + "learning_rate": 1.4529613407721192e-05, + "loss": 0.2016, + "step": 4630 + }, + { + "epoch": 2.024017467248908, + "grad_norm": 0.21939396929192034, + "learning_rate": 1.4471973185873888e-05, + "loss": 0.2092, + "step": 4635 + }, + { + "epoch": 2.0262008733624453, + "grad_norm": 0.20467895973459885, + "learning_rate": 1.4414400913799535e-05, + "loss": 0.2026, + "step": 4640 + }, + { + "epoch": 2.0283842794759823, + "grad_norm": 0.20303046431819063, + "learning_rate": 1.4356896963079885e-05, + "loss": 0.2034, + "step": 4645 + }, + { + "epoch": 2.03056768558952, + "grad_norm": 0.2147257371596202, + "learning_rate": 1.4299461704855759e-05, + "loss": 0.2038, + "step": 4650 + }, + { + "epoch": 2.032751091703057, + "grad_norm": 0.2709689872203016, + "learning_rate": 1.424209550982461e-05, + "loss": 0.2151, + "step": 4655 + }, + { + "epoch": 2.034934497816594, + "grad_norm": 0.22708728340724726, + "learning_rate": 1.4184798748238148e-05, + "loss": 0.2176, + "step": 4660 + }, + { + "epoch": 2.037117903930131, + "grad_norm": 0.25203147090946515, + "learning_rate": 1.4127571789899946e-05, + "loss": 0.2077, + "step": 4665 + }, + { + "epoch": 2.039301310043668, + "grad_norm": 0.23983946431243502, + "learning_rate": 1.4070415004163056e-05, + "loss": 0.2163, + "step": 4670 + }, + { + "epoch": 2.041484716157205, + "grad_norm": 0.24658657899765385, + "learning_rate": 1.4013328759927625e-05, + "loss": 0.2139, + "step": 4675 + }, + { + "epoch": 2.0436681222707422, + "grad_norm": 0.22543241349761747, + "learning_rate": 1.39563134256385e-05, + "loss": 0.2019, + "step": 4680 + }, + { + "epoch": 2.0458515283842793, + "grad_norm": 0.22236910097875634, + "learning_rate": 1.3899369369282888e-05, + "loss": 0.2183, + "step": 4685 + }, + { + "epoch": 2.0480349344978164, + "grad_norm": 0.20505749283945437, + "learning_rate": 1.3842496958387938e-05, + "loss": 0.2188, + "step": 4690 + }, + { + "epoch": 2.050218340611354, + "grad_norm": 0.2025603121212584, + "learning_rate": 1.3785696560018374e-05, + "loss": 0.2038, + "step": 4695 + }, + { + "epoch": 2.052401746724891, + "grad_norm": 0.2263051334719587, + "learning_rate": 1.372896854077416e-05, + "loss": 0.206, + "step": 4700 + }, + { + "epoch": 2.054585152838428, + "grad_norm": 0.2321594137504455, + "learning_rate": 1.3672313266788106e-05, + "loss": 0.2133, + "step": 4705 + }, + { + "epoch": 2.056768558951965, + "grad_norm": 0.2122041960683085, + "learning_rate": 1.3615731103723507e-05, + "loss": 0.2126, + "step": 4710 + }, + { + "epoch": 2.058951965065502, + "grad_norm": 0.24859114020141224, + "learning_rate": 1.3559222416771793e-05, + "loss": 0.2061, + "step": 4715 + }, + { + "epoch": 2.061135371179039, + "grad_norm": 0.21853834679540102, + "learning_rate": 1.350278757065016e-05, + "loss": 0.2109, + "step": 4720 + }, + { + "epoch": 2.0633187772925763, + "grad_norm": 0.2074388039569142, + "learning_rate": 1.3446426929599229e-05, + "loss": 0.2137, + "step": 4725 + }, + { + "epoch": 2.0655021834061134, + "grad_norm": 0.21863003094886263, + "learning_rate": 1.3390140857380678e-05, + "loss": 0.1967, + "step": 4730 + }, + { + "epoch": 2.067685589519651, + "grad_norm": 0.25260685298764257, + "learning_rate": 1.3333929717274919e-05, + "loss": 0.2255, + "step": 4735 + }, + { + "epoch": 2.069868995633188, + "grad_norm": 0.23960033740421333, + "learning_rate": 1.3277793872078732e-05, + "loss": 0.2195, + "step": 4740 + }, + { + "epoch": 2.072052401746725, + "grad_norm": 0.23761525700313252, + "learning_rate": 1.3221733684102905e-05, + "loss": 0.2128, + "step": 4745 + }, + { + "epoch": 2.074235807860262, + "grad_norm": 0.21987521365657883, + "learning_rate": 1.3165749515169967e-05, + "loss": 0.2115, + "step": 4750 + }, + { + "epoch": 2.076419213973799, + "grad_norm": 0.223938388935573, + "learning_rate": 1.310984172661177e-05, + "loss": 0.2181, + "step": 4755 + }, + { + "epoch": 2.078602620087336, + "grad_norm": 0.2521445985500392, + "learning_rate": 1.305401067926722e-05, + "loss": 0.2143, + "step": 4760 + }, + { + "epoch": 2.0807860262008733, + "grad_norm": 0.23287630698929376, + "learning_rate": 1.2998256733479897e-05, + "loss": 0.2002, + "step": 4765 + }, + { + "epoch": 2.0829694323144103, + "grad_norm": 0.2600254334217533, + "learning_rate": 1.294258024909577e-05, + "loss": 0.2041, + "step": 4770 + }, + { + "epoch": 2.0851528384279474, + "grad_norm": 0.2926480056967554, + "learning_rate": 1.288698158546085e-05, + "loss": 0.2206, + "step": 4775 + }, + { + "epoch": 2.087336244541485, + "grad_norm": 0.21634943547113875, + "learning_rate": 1.2831461101418884e-05, + "loss": 0.2166, + "step": 4780 + }, + { + "epoch": 2.089519650655022, + "grad_norm": 0.2398092797244653, + "learning_rate": 1.2776019155309035e-05, + "loss": 0.2228, + "step": 4785 + }, + { + "epoch": 2.091703056768559, + "grad_norm": 0.22191669799449495, + "learning_rate": 1.2720656104963569e-05, + "loss": 0.2146, + "step": 4790 + }, + { + "epoch": 2.093886462882096, + "grad_norm": 0.22249356654838032, + "learning_rate": 1.2665372307705525e-05, + "loss": 0.2173, + "step": 4795 + }, + { + "epoch": 2.096069868995633, + "grad_norm": 0.2338757269866602, + "learning_rate": 1.2610168120346455e-05, + "loss": 0.2203, + "step": 4800 + }, + { + "epoch": 2.0982532751091703, + "grad_norm": 0.20569835239307332, + "learning_rate": 1.2555043899184077e-05, + "loss": 0.2169, + "step": 4805 + }, + { + "epoch": 2.1004366812227073, + "grad_norm": 0.18840672821070764, + "learning_rate": 1.2500000000000006e-05, + "loss": 0.2131, + "step": 4810 + }, + { + "epoch": 2.1026200873362444, + "grad_norm": 0.229982583270256, + "learning_rate": 1.2445036778057434e-05, + "loss": 0.2155, + "step": 4815 + }, + { + "epoch": 2.1048034934497815, + "grad_norm": 0.21678928738468423, + "learning_rate": 1.2390154588098854e-05, + "loss": 0.2151, + "step": 4820 + }, + { + "epoch": 2.106986899563319, + "grad_norm": 0.20815591769350697, + "learning_rate": 1.233535378434376e-05, + "loss": 0.2075, + "step": 4825 + }, + { + "epoch": 2.109170305676856, + "grad_norm": 0.22153362078682773, + "learning_rate": 1.2280634720486364e-05, + "loss": 0.2136, + "step": 4830 + }, + { + "epoch": 2.111353711790393, + "grad_norm": 0.21312931677811298, + "learning_rate": 1.2225997749693318e-05, + "loss": 0.21, + "step": 4835 + }, + { + "epoch": 2.11353711790393, + "grad_norm": 0.23940821453611505, + "learning_rate": 1.2171443224601443e-05, + "loss": 0.2141, + "step": 4840 + }, + { + "epoch": 2.1157205240174672, + "grad_norm": 0.2242248154195571, + "learning_rate": 1.2116971497315402e-05, + "loss": 0.208, + "step": 4845 + }, + { + "epoch": 2.1179039301310043, + "grad_norm": 0.21221180992392646, + "learning_rate": 1.2062582919405502e-05, + "loss": 0.2093, + "step": 4850 + }, + { + "epoch": 2.1200873362445414, + "grad_norm": 0.22445864486046593, + "learning_rate": 1.200827784190537e-05, + "loss": 0.2146, + "step": 4855 + }, + { + "epoch": 2.1222707423580784, + "grad_norm": 0.2301722383710042, + "learning_rate": 1.1954056615309728e-05, + "loss": 0.2084, + "step": 4860 + }, + { + "epoch": 2.1244541484716155, + "grad_norm": 0.2135558798950853, + "learning_rate": 1.1899919589572092e-05, + "loss": 0.2126, + "step": 4865 + }, + { + "epoch": 2.126637554585153, + "grad_norm": 0.2073308539610393, + "learning_rate": 1.1845867114102532e-05, + "loss": 0.2063, + "step": 4870 + }, + { + "epoch": 2.12882096069869, + "grad_norm": 0.2049312257048255, + "learning_rate": 1.1791899537765416e-05, + "loss": 0.2224, + "step": 4875 + }, + { + "epoch": 2.131004366812227, + "grad_norm": 0.24321581226405323, + "learning_rate": 1.173801720887717e-05, + "loss": 0.2218, + "step": 4880 + }, + { + "epoch": 2.1331877729257642, + "grad_norm": 0.21578319625044598, + "learning_rate": 1.1684220475203986e-05, + "loss": 0.199, + "step": 4885 + }, + { + "epoch": 2.1353711790393013, + "grad_norm": 0.2162898287189938, + "learning_rate": 1.1630509683959643e-05, + "loss": 0.2121, + "step": 4890 + }, + { + "epoch": 2.1375545851528384, + "grad_norm": 0.24512267530291906, + "learning_rate": 1.1576885181803215e-05, + "loss": 0.2171, + "step": 4895 + }, + { + "epoch": 2.1397379912663754, + "grad_norm": 0.19666073074843837, + "learning_rate": 1.1523347314836858e-05, + "loss": 0.2135, + "step": 4900 + }, + { + "epoch": 2.1419213973799125, + "grad_norm": 0.21309169492333768, + "learning_rate": 1.1469896428603563e-05, + "loss": 0.22, + "step": 4905 + }, + { + "epoch": 2.14410480349345, + "grad_norm": 0.2097865812131667, + "learning_rate": 1.1416532868084937e-05, + "loss": 0.2022, + "step": 4910 + }, + { + "epoch": 2.146288209606987, + "grad_norm": 0.22140108936049668, + "learning_rate": 1.1363256977698968e-05, + "loss": 0.2157, + "step": 4915 + }, + { + "epoch": 2.148471615720524, + "grad_norm": 0.20612571861461668, + "learning_rate": 1.1310069101297807e-05, + "loss": 0.212, + "step": 4920 + }, + { + "epoch": 2.150655021834061, + "grad_norm": 0.2172078322760656, + "learning_rate": 1.1256969582165544e-05, + "loss": 0.21, + "step": 4925 + }, + { + "epoch": 2.1528384279475983, + "grad_norm": 0.22655837718087724, + "learning_rate": 1.1203958763016007e-05, + "loss": 0.1969, + "step": 4930 + }, + { + "epoch": 2.1550218340611353, + "grad_norm": 0.23562847936093284, + "learning_rate": 1.1151036985990509e-05, + "loss": 0.209, + "step": 4935 + }, + { + "epoch": 2.1572052401746724, + "grad_norm": 0.21438898804486692, + "learning_rate": 1.1098204592655703e-05, + "loss": 0.2144, + "step": 4940 + }, + { + "epoch": 2.1593886462882095, + "grad_norm": 0.22341501985465956, + "learning_rate": 1.1045461924001325e-05, + "loss": 0.2095, + "step": 4945 + }, + { + "epoch": 2.1615720524017465, + "grad_norm": 0.2154153191432171, + "learning_rate": 1.0992809320438016e-05, + "loss": 0.2071, + "step": 4950 + }, + { + "epoch": 2.1637554585152836, + "grad_norm": 0.19531686390702627, + "learning_rate": 1.0940247121795123e-05, + "loss": 0.2095, + "step": 4955 + }, + { + "epoch": 2.165938864628821, + "grad_norm": 0.22132766807603957, + "learning_rate": 1.08877756673185e-05, + "loss": 0.2174, + "step": 4960 + }, + { + "epoch": 2.168122270742358, + "grad_norm": 0.22517823657634206, + "learning_rate": 1.0835395295668327e-05, + "loss": 0.2179, + "step": 4965 + }, + { + "epoch": 2.1703056768558953, + "grad_norm": 0.21142342412266563, + "learning_rate": 1.0783106344916913e-05, + "loss": 0.2051, + "step": 4970 + }, + { + "epoch": 2.1724890829694323, + "grad_norm": 0.20941158976258883, + "learning_rate": 1.073090915254652e-05, + "loss": 0.2221, + "step": 4975 + }, + { + "epoch": 2.1746724890829694, + "grad_norm": 0.21006000563282354, + "learning_rate": 1.06788040554472e-05, + "loss": 0.2003, + "step": 4980 + }, + { + "epoch": 2.1768558951965065, + "grad_norm": 0.19371851293041617, + "learning_rate": 1.0626791389914573e-05, + "loss": 0.2029, + "step": 4985 + }, + { + "epoch": 2.1790393013100435, + "grad_norm": 0.2202294387310257, + "learning_rate": 1.0574871491647723e-05, + "loss": 0.2141, + "step": 4990 + }, + { + "epoch": 2.1812227074235806, + "grad_norm": 0.22059840659575228, + "learning_rate": 1.0523044695746975e-05, + "loss": 0.2121, + "step": 4995 + }, + { + "epoch": 2.183406113537118, + "grad_norm": 0.2066703461639902, + "learning_rate": 1.0471311336711772e-05, + "loss": 0.2123, + "step": 5000 + }, + { + "epoch": 2.185589519650655, + "grad_norm": 0.20919147411357605, + "learning_rate": 1.0419671748438487e-05, + "loss": 0.2166, + "step": 5005 + }, + { + "epoch": 2.1877729257641922, + "grad_norm": 0.1934804236700962, + "learning_rate": 1.0368126264218282e-05, + "loss": 0.2109, + "step": 5010 + }, + { + "epoch": 2.1899563318777293, + "grad_norm": 0.2089470879409506, + "learning_rate": 1.0316675216734952e-05, + "loss": 0.2057, + "step": 5015 + }, + { + "epoch": 2.1921397379912664, + "grad_norm": 0.19986998684970525, + "learning_rate": 1.026531893806279e-05, + "loss": 0.2225, + "step": 5020 + }, + { + "epoch": 2.1943231441048034, + "grad_norm": 0.22134729099448455, + "learning_rate": 1.0214057759664424e-05, + "loss": 0.2161, + "step": 5025 + }, + { + "epoch": 2.1965065502183405, + "grad_norm": 0.20179325283452076, + "learning_rate": 1.0162892012388703e-05, + "loss": 0.223, + "step": 5030 + }, + { + "epoch": 2.1986899563318776, + "grad_norm": 0.20653676340802546, + "learning_rate": 1.0111822026468515e-05, + "loss": 0.2127, + "step": 5035 + }, + { + "epoch": 2.2008733624454146, + "grad_norm": 0.20118347192898964, + "learning_rate": 1.0060848131518716e-05, + "loss": 0.2132, + "step": 5040 + }, + { + "epoch": 2.203056768558952, + "grad_norm": 0.2127487431663625, + "learning_rate": 1.0009970656533965e-05, + "loss": 0.2138, + "step": 5045 + }, + { + "epoch": 2.2052401746724892, + "grad_norm": 0.18873587826004806, + "learning_rate": 9.95918992988661e-06, + "loss": 0.2159, + "step": 5050 + }, + { + "epoch": 2.2074235807860263, + "grad_norm": 0.21267677980729546, + "learning_rate": 9.908506279324564e-06, + "loss": 0.2094, + "step": 5055 + }, + { + "epoch": 2.2096069868995634, + "grad_norm": 0.20619431733016733, + "learning_rate": 9.857920031969198e-06, + "loss": 0.2066, + "step": 5060 + }, + { + "epoch": 2.2117903930131004, + "grad_norm": 0.2223339638565872, + "learning_rate": 9.807431514313225e-06, + "loss": 0.2315, + "step": 5065 + }, + { + "epoch": 2.2139737991266375, + "grad_norm": 0.2351717184649456, + "learning_rate": 9.757041052218587e-06, + "loss": 0.2192, + "step": 5070 + }, + { + "epoch": 2.2161572052401746, + "grad_norm": 0.18769468217068455, + "learning_rate": 9.70674897091437e-06, + "loss": 0.21, + "step": 5075 + }, + { + "epoch": 2.2183406113537116, + "grad_norm": 0.24549048829311135, + "learning_rate": 9.656555594994687e-06, + "loss": 0.2025, + "step": 5080 + }, + { + "epoch": 2.2205240174672487, + "grad_norm": 0.20415216492897895, + "learning_rate": 9.606461248416573e-06, + "loss": 0.2016, + "step": 5085 + }, + { + "epoch": 2.222707423580786, + "grad_norm": 0.2005430180173854, + "learning_rate": 9.556466254497931e-06, + "loss": 0.2098, + "step": 5090 + }, + { + "epoch": 2.2248908296943233, + "grad_norm": 0.22132570053896475, + "learning_rate": 9.506570935915419e-06, + "loss": 0.2087, + "step": 5095 + }, + { + "epoch": 2.2270742358078603, + "grad_norm": 0.19989818333839046, + "learning_rate": 9.456775614702364e-06, + "loss": 0.2079, + "step": 5100 + }, + { + "epoch": 2.2292576419213974, + "grad_norm": 0.2473690640544336, + "learning_rate": 9.407080612246697e-06, + "loss": 0.2107, + "step": 5105 + }, + { + "epoch": 2.2314410480349345, + "grad_norm": 0.21359326359845074, + "learning_rate": 9.357486249288891e-06, + "loss": 0.2254, + "step": 5110 + }, + { + "epoch": 2.2336244541484715, + "grad_norm": 0.2355252231386045, + "learning_rate": 9.307992845919844e-06, + "loss": 0.2032, + "step": 5115 + }, + { + "epoch": 2.2358078602620086, + "grad_norm": 0.1990769526625258, + "learning_rate": 9.258600721578872e-06, + "loss": 0.2076, + "step": 5120 + }, + { + "epoch": 2.2379912663755457, + "grad_norm": 0.2244729672302675, + "learning_rate": 9.209310195051582e-06, + "loss": 0.2186, + "step": 5125 + }, + { + "epoch": 2.2401746724890828, + "grad_norm": 0.19571847441525664, + "learning_rate": 9.160121584467878e-06, + "loss": 0.1942, + "step": 5130 + }, + { + "epoch": 2.2423580786026203, + "grad_norm": 0.2174940143142323, + "learning_rate": 9.111035207299878e-06, + "loss": 0.2155, + "step": 5135 + }, + { + "epoch": 2.2445414847161573, + "grad_norm": 0.20301405504923226, + "learning_rate": 9.062051380359857e-06, + "loss": 0.2155, + "step": 5140 + }, + { + "epoch": 2.2467248908296944, + "grad_norm": 0.2381569949311249, + "learning_rate": 9.01317041979822e-06, + "loss": 0.2191, + "step": 5145 + }, + { + "epoch": 2.2489082969432315, + "grad_norm": 0.20675760421426403, + "learning_rate": 8.964392641101455e-06, + "loss": 0.198, + "step": 5150 + }, + { + "epoch": 2.2510917030567685, + "grad_norm": 0.18783552296846562, + "learning_rate": 8.915718359090086e-06, + "loss": 0.2049, + "step": 5155 + }, + { + "epoch": 2.2532751091703056, + "grad_norm": 0.20565919831917942, + "learning_rate": 8.867147887916663e-06, + "loss": 0.2227, + "step": 5160 + }, + { + "epoch": 2.2554585152838427, + "grad_norm": 0.205583376077084, + "learning_rate": 8.818681541063714e-06, + "loss": 0.2118, + "step": 5165 + }, + { + "epoch": 2.2576419213973797, + "grad_norm": 0.18747729101469687, + "learning_rate": 8.770319631341745e-06, + "loss": 0.2014, + "step": 5170 + }, + { + "epoch": 2.2598253275109172, + "grad_norm": 0.1903688682819849, + "learning_rate": 8.722062470887169e-06, + "loss": 0.1973, + "step": 5175 + }, + { + "epoch": 2.2620087336244543, + "grad_norm": 0.20544378450917616, + "learning_rate": 8.673910371160368e-06, + "loss": 0.2132, + "step": 5180 + }, + { + "epoch": 2.2641921397379914, + "grad_norm": 0.22300030961664347, + "learning_rate": 8.625863642943623e-06, + "loss": 0.2102, + "step": 5185 + }, + { + "epoch": 2.2663755458515285, + "grad_norm": 0.24719313551585811, + "learning_rate": 8.577922596339142e-06, + "loss": 0.2091, + "step": 5190 + }, + { + "epoch": 2.2685589519650655, + "grad_norm": 0.19395953090113746, + "learning_rate": 8.530087540767029e-06, + "loss": 0.2119, + "step": 5195 + }, + { + "epoch": 2.2707423580786026, + "grad_norm": 0.2057558795183291, + "learning_rate": 8.482358784963312e-06, + "loss": 0.2088, + "step": 5200 + }, + { + "epoch": 2.2729257641921397, + "grad_norm": 0.194298675532194, + "learning_rate": 8.43473663697794e-06, + "loss": 0.1949, + "step": 5205 + }, + { + "epoch": 2.2751091703056767, + "grad_norm": 0.2059192410189242, + "learning_rate": 8.387221404172796e-06, + "loss": 0.2036, + "step": 5210 + }, + { + "epoch": 2.277292576419214, + "grad_norm": 0.20744243911102733, + "learning_rate": 8.339813393219714e-06, + "loss": 0.2146, + "step": 5215 + }, + { + "epoch": 2.279475982532751, + "grad_norm": 0.2043265849228992, + "learning_rate": 8.292512910098508e-06, + "loss": 0.2206, + "step": 5220 + }, + { + "epoch": 2.2816593886462884, + "grad_norm": 0.2100815969871642, + "learning_rate": 8.245320260094955e-06, + "loss": 0.2199, + "step": 5225 + }, + { + "epoch": 2.2838427947598254, + "grad_norm": 0.20976956534920874, + "learning_rate": 8.198235747798893e-06, + "loss": 0.2168, + "step": 5230 + }, + { + "epoch": 2.2860262008733625, + "grad_norm": 0.2050232275697135, + "learning_rate": 8.151259677102211e-06, + "loss": 0.216, + "step": 5235 + }, + { + "epoch": 2.2882096069868996, + "grad_norm": 4.122598477737714, + "learning_rate": 8.104392351196885e-06, + "loss": 0.2198, + "step": 5240 + }, + { + "epoch": 2.2903930131004366, + "grad_norm": 0.2237909328728828, + "learning_rate": 8.057634072573048e-06, + "loss": 0.2173, + "step": 5245 + }, + { + "epoch": 2.2925764192139737, + "grad_norm": 0.21176821690953787, + "learning_rate": 8.010985143017008e-06, + "loss": 0.2157, + "step": 5250 + }, + { + "epoch": 2.2947598253275108, + "grad_norm": 0.8545824488764707, + "learning_rate": 7.964445863609329e-06, + "loss": 0.215, + "step": 5255 + }, + { + "epoch": 2.2969432314410483, + "grad_norm": 0.23797387294881264, + "learning_rate": 7.91801653472286e-06, + "loss": 0.1993, + "step": 5260 + }, + { + "epoch": 2.2991266375545854, + "grad_norm": 0.209126744021443, + "learning_rate": 7.871697456020819e-06, + "loss": 0.2285, + "step": 5265 + }, + { + "epoch": 2.3013100436681224, + "grad_norm": 0.2036921534501079, + "learning_rate": 7.825488926454851e-06, + "loss": 0.2137, + "step": 5270 + }, + { + "epoch": 2.3034934497816595, + "grad_norm": 0.19891864557452632, + "learning_rate": 7.77939124426308e-06, + "loss": 0.2094, + "step": 5275 + }, + { + "epoch": 2.3056768558951966, + "grad_norm": 0.21772742093397335, + "learning_rate": 7.733404706968215e-06, + "loss": 0.2097, + "step": 5280 + }, + { + "epoch": 2.3078602620087336, + "grad_norm": 0.1920838796289188, + "learning_rate": 7.687529611375616e-06, + "loss": 0.2161, + "step": 5285 + }, + { + "epoch": 2.3100436681222707, + "grad_norm": 0.20114220947132505, + "learning_rate": 7.64176625357138e-06, + "loss": 0.2041, + "step": 5290 + }, + { + "epoch": 2.3122270742358078, + "grad_norm": 0.20058172964283594, + "learning_rate": 7.596114928920422e-06, + "loss": 0.2124, + "step": 5295 + }, + { + "epoch": 2.314410480349345, + "grad_norm": 0.20110764788896288, + "learning_rate": 7.55057593206458e-06, + "loss": 0.2018, + "step": 5300 + }, + { + "epoch": 2.316593886462882, + "grad_norm": 0.2033865906846183, + "learning_rate": 7.505149556920699e-06, + "loss": 0.2139, + "step": 5305 + }, + { + "epoch": 2.3187772925764194, + "grad_norm": 0.21910248265310586, + "learning_rate": 7.459836096678757e-06, + "loss": 0.2153, + "step": 5310 + }, + { + "epoch": 2.3209606986899565, + "grad_norm": 0.2026773384631961, + "learning_rate": 7.41463584379995e-06, + "loss": 0.215, + "step": 5315 + }, + { + "epoch": 2.3231441048034935, + "grad_norm": 0.2127225287015194, + "learning_rate": 7.369549090014821e-06, + "loss": 0.212, + "step": 5320 + }, + { + "epoch": 2.3253275109170306, + "grad_norm": 0.20869494683552092, + "learning_rate": 7.32457612632135e-06, + "loss": 0.216, + "step": 5325 + }, + { + "epoch": 2.3275109170305677, + "grad_norm": 0.19166630407586727, + "learning_rate": 7.279717242983117e-06, + "loss": 0.2166, + "step": 5330 + }, + { + "epoch": 2.3296943231441047, + "grad_norm": 0.2004851412037729, + "learning_rate": 7.234972729527401e-06, + "loss": 0.207, + "step": 5335 + }, + { + "epoch": 2.331877729257642, + "grad_norm": 0.18554259866541592, + "learning_rate": 7.190342874743311e-06, + "loss": 0.201, + "step": 5340 + }, + { + "epoch": 2.334061135371179, + "grad_norm": 0.21209107877243255, + "learning_rate": 7.145827966679936e-06, + "loss": 0.1968, + "step": 5345 + }, + { + "epoch": 2.3362445414847164, + "grad_norm": 0.20153718074405388, + "learning_rate": 7.101428292644477e-06, + "loss": 0.2176, + "step": 5350 + }, + { + "epoch": 2.3384279475982535, + "grad_norm": 0.18992678255749726, + "learning_rate": 7.057144139200378e-06, + "loss": 0.2013, + "step": 5355 + }, + { + "epoch": 2.3406113537117905, + "grad_norm": 0.1830289392115265, + "learning_rate": 7.0129757921655355e-06, + "loss": 0.1979, + "step": 5360 + }, + { + "epoch": 2.3427947598253276, + "grad_norm": 0.19740674215886508, + "learning_rate": 6.968923536610356e-06, + "loss": 0.2078, + "step": 5365 + }, + { + "epoch": 2.3449781659388647, + "grad_norm": 0.2121990875989476, + "learning_rate": 6.924987656856002e-06, + "loss": 0.1936, + "step": 5370 + }, + { + "epoch": 2.3471615720524017, + "grad_norm": 0.22075994643117536, + "learning_rate": 6.881168436472524e-06, + "loss": 0.2114, + "step": 5375 + }, + { + "epoch": 2.349344978165939, + "grad_norm": 0.20805940844514723, + "learning_rate": 6.837466158277025e-06, + "loss": 0.1941, + "step": 5380 + }, + { + "epoch": 2.351528384279476, + "grad_norm": 0.18880443913500422, + "learning_rate": 6.7938811043318475e-06, + "loss": 0.2151, + "step": 5385 + }, + { + "epoch": 2.353711790393013, + "grad_norm": 0.23050444783405694, + "learning_rate": 6.750413555942742e-06, + "loss": 0.2066, + "step": 5390 + }, + { + "epoch": 2.35589519650655, + "grad_norm": 0.19659728337288793, + "learning_rate": 6.707063793657065e-06, + "loss": 0.2168, + "step": 5395 + }, + { + "epoch": 2.3580786026200875, + "grad_norm": 0.2044183886617955, + "learning_rate": 6.663832097261952e-06, + "loss": 0.2193, + "step": 5400 + }, + { + "epoch": 2.3602620087336246, + "grad_norm": 0.19473776562391912, + "learning_rate": 6.620718745782525e-06, + "loss": 0.2031, + "step": 5405 + }, + { + "epoch": 2.3624454148471616, + "grad_norm": 0.19433516742290474, + "learning_rate": 6.577724017480094e-06, + "loss": 0.2147, + "step": 5410 + }, + { + "epoch": 2.3646288209606987, + "grad_norm": 0.2002897275632141, + "learning_rate": 6.534848189850329e-06, + "loss": 0.1992, + "step": 5415 + }, + { + "epoch": 2.3668122270742358, + "grad_norm": 0.1981020777647114, + "learning_rate": 6.492091539621517e-06, + "loss": 0.2029, + "step": 5420 + }, + { + "epoch": 2.368995633187773, + "grad_norm": 0.18263004359615023, + "learning_rate": 6.449454342752748e-06, + "loss": 0.2018, + "step": 5425 + }, + { + "epoch": 2.37117903930131, + "grad_norm": 0.1878131474845262, + "learning_rate": 6.406936874432134e-06, + "loss": 0.2037, + "step": 5430 + }, + { + "epoch": 2.373362445414847, + "grad_norm": 0.204062897390976, + "learning_rate": 6.364539409075043e-06, + "loss": 0.2017, + "step": 5435 + }, + { + "epoch": 2.3755458515283845, + "grad_norm": 0.19374315646095067, + "learning_rate": 6.322262220322314e-06, + "loss": 0.2093, + "step": 5440 + }, + { + "epoch": 2.3777292576419216, + "grad_norm": 0.20250412793089098, + "learning_rate": 6.280105581038509e-06, + "loss": 0.2031, + "step": 5445 + }, + { + "epoch": 2.3799126637554586, + "grad_norm": 0.18639353569932318, + "learning_rate": 6.238069763310136e-06, + "loss": 0.2082, + "step": 5450 + }, + { + "epoch": 2.3820960698689957, + "grad_norm": 0.19128091651488877, + "learning_rate": 6.1961550384438984e-06, + "loss": 0.2066, + "step": 5455 + }, + { + "epoch": 2.3842794759825328, + "grad_norm": 0.19608351533598506, + "learning_rate": 6.154361676964956e-06, + "loss": 0.2165, + "step": 5460 + }, + { + "epoch": 2.38646288209607, + "grad_norm": 0.18656964338831433, + "learning_rate": 6.112689948615144e-06, + "loss": 0.2019, + "step": 5465 + }, + { + "epoch": 2.388646288209607, + "grad_norm": 0.18488662819515192, + "learning_rate": 6.071140122351276e-06, + "loss": 0.2018, + "step": 5470 + }, + { + "epoch": 2.390829694323144, + "grad_norm": 0.19839674782985758, + "learning_rate": 6.029712466343382e-06, + "loss": 0.2057, + "step": 5475 + }, + { + "epoch": 2.393013100436681, + "grad_norm": 0.1935316894500808, + "learning_rate": 5.988407247972982e-06, + "loss": 0.2101, + "step": 5480 + }, + { + "epoch": 2.395196506550218, + "grad_norm": 0.19014181610486888, + "learning_rate": 5.947224733831364e-06, + "loss": 0.214, + "step": 5485 + }, + { + "epoch": 2.3973799126637556, + "grad_norm": 0.2047537532894175, + "learning_rate": 5.906165189717855e-06, + "loss": 0.2069, + "step": 5490 + }, + { + "epoch": 2.3995633187772927, + "grad_norm": 0.20244514937691652, + "learning_rate": 5.8652288806381215e-06, + "loss": 0.2093, + "step": 5495 + }, + { + "epoch": 2.4017467248908297, + "grad_norm": 0.2029682194929424, + "learning_rate": 5.824416070802438e-06, + "loss": 0.1989, + "step": 5500 + }, + { + "epoch": 2.403930131004367, + "grad_norm": 0.19862209705141703, + "learning_rate": 5.7837270236239995e-06, + "loss": 0.1988, + "step": 5505 + }, + { + "epoch": 2.406113537117904, + "grad_norm": 0.19816634806985825, + "learning_rate": 5.74316200171722e-06, + "loss": 0.2042, + "step": 5510 + }, + { + "epoch": 2.408296943231441, + "grad_norm": 0.20129123615306388, + "learning_rate": 5.70272126689601e-06, + "loss": 0.2036, + "step": 5515 + }, + { + "epoch": 2.410480349344978, + "grad_norm": 0.18987017520409014, + "learning_rate": 5.662405080172134e-06, + "loss": 0.2078, + "step": 5520 + }, + { + "epoch": 2.4126637554585155, + "grad_norm": 0.19052375793050977, + "learning_rate": 5.622213701753487e-06, + "loss": 0.2231, + "step": 5525 + }, + { + "epoch": 2.4148471615720526, + "grad_norm": 0.19323234507341908, + "learning_rate": 5.582147391042433e-06, + "loss": 0.2092, + "step": 5530 + }, + { + "epoch": 2.4170305676855897, + "grad_norm": 0.1924708259179898, + "learning_rate": 5.542206406634129e-06, + "loss": 0.2078, + "step": 5535 + }, + { + "epoch": 2.4192139737991267, + "grad_norm": 0.207006675227268, + "learning_rate": 5.5023910063148445e-06, + "loss": 0.2109, + "step": 5540 + }, + { + "epoch": 2.421397379912664, + "grad_norm": 0.2002952114074735, + "learning_rate": 5.46270144706032e-06, + "loss": 0.198, + "step": 5545 + }, + { + "epoch": 2.423580786026201, + "grad_norm": 0.2091856914324174, + "learning_rate": 5.423137985034085e-06, + "loss": 0.2138, + "step": 5550 + }, + { + "epoch": 2.425764192139738, + "grad_norm": 0.22941186423542465, + "learning_rate": 5.3837008755858145e-06, + "loss": 0.2013, + "step": 5555 + }, + { + "epoch": 2.427947598253275, + "grad_norm": 0.1974516927188841, + "learning_rate": 5.3443903732496966e-06, + "loss": 0.2069, + "step": 5560 + }, + { + "epoch": 2.430131004366812, + "grad_norm": 0.1861177274579643, + "learning_rate": 5.305206731742748e-06, + "loss": 0.2086, + "step": 5565 + }, + { + "epoch": 2.432314410480349, + "grad_norm": 0.196399769855382, + "learning_rate": 5.266150203963224e-06, + "loss": 0.2155, + "step": 5570 + }, + { + "epoch": 2.4344978165938866, + "grad_norm": 0.20583463409268135, + "learning_rate": 5.227221041988956e-06, + "loss": 0.2055, + "step": 5575 + }, + { + "epoch": 2.4366812227074237, + "grad_norm": 0.19145217163247738, + "learning_rate": 5.188419497075739e-06, + "loss": 0.2045, + "step": 5580 + }, + { + "epoch": 2.438864628820961, + "grad_norm": 0.2300625133365478, + "learning_rate": 5.1497458196556995e-06, + "loss": 0.2192, + "step": 5585 + }, + { + "epoch": 2.441048034934498, + "grad_norm": 0.19252331533699785, + "learning_rate": 5.111200259335688e-06, + "loss": 0.2089, + "step": 5590 + }, + { + "epoch": 2.443231441048035, + "grad_norm": 0.20014832340012859, + "learning_rate": 5.072783064895664e-06, + "loss": 0.2243, + "step": 5595 + }, + { + "epoch": 2.445414847161572, + "grad_norm": 0.19507458711265246, + "learning_rate": 5.034494484287086e-06, + "loss": 0.2195, + "step": 5600 + }, + { + "epoch": 2.447598253275109, + "grad_norm": 0.18618941255923363, + "learning_rate": 4.996334764631322e-06, + "loss": 0.2034, + "step": 5605 + }, + { + "epoch": 2.449781659388646, + "grad_norm": 0.2223628562088218, + "learning_rate": 4.958304152218046e-06, + "loss": 0.2076, + "step": 5610 + }, + { + "epoch": 2.4519650655021836, + "grad_norm": 0.22749716704810904, + "learning_rate": 4.920402892503648e-06, + "loss": 0.2142, + "step": 5615 + }, + { + "epoch": 2.4541484716157207, + "grad_norm": 0.20326433502448837, + "learning_rate": 4.882631230109655e-06, + "loss": 0.206, + "step": 5620 + }, + { + "epoch": 2.4563318777292578, + "grad_norm": 0.20525401555309347, + "learning_rate": 4.844989408821149e-06, + "loss": 0.2042, + "step": 5625 + }, + { + "epoch": 2.458515283842795, + "grad_norm": 0.20210008215232908, + "learning_rate": 4.807477671585195e-06, + "loss": 0.1999, + "step": 5630 + }, + { + "epoch": 2.460698689956332, + "grad_norm": 0.21710829320874975, + "learning_rate": 4.770096260509274e-06, + "loss": 0.203, + "step": 5635 + }, + { + "epoch": 2.462882096069869, + "grad_norm": 0.18303629687950113, + "learning_rate": 4.732845416859713e-06, + "loss": 0.2134, + "step": 5640 + }, + { + "epoch": 2.465065502183406, + "grad_norm": 0.19284778474305014, + "learning_rate": 4.695725381060137e-06, + "loss": 0.2062, + "step": 5645 + }, + { + "epoch": 2.467248908296943, + "grad_norm": 0.18825083490369524, + "learning_rate": 4.658736392689922e-06, + "loss": 0.2214, + "step": 5650 + }, + { + "epoch": 2.46943231441048, + "grad_norm": 0.20310843234802933, + "learning_rate": 4.6218786904826175e-06, + "loss": 0.2132, + "step": 5655 + }, + { + "epoch": 2.4716157205240172, + "grad_norm": 0.21879459935317394, + "learning_rate": 4.58515251232445e-06, + "loss": 0.2193, + "step": 5660 + }, + { + "epoch": 2.4737991266375547, + "grad_norm": 0.2064948920674535, + "learning_rate": 4.548558095252758e-06, + "loss": 0.206, + "step": 5665 + }, + { + "epoch": 2.475982532751092, + "grad_norm": 0.20819617169964372, + "learning_rate": 4.512095675454472e-06, + "loss": 0.2116, + "step": 5670 + }, + { + "epoch": 2.478165938864629, + "grad_norm": 0.19013108098717665, + "learning_rate": 4.475765488264588e-06, + "loss": 0.2031, + "step": 5675 + }, + { + "epoch": 2.480349344978166, + "grad_norm": 0.19875013946254505, + "learning_rate": 4.439567768164651e-06, + "loss": 0.2017, + "step": 5680 + }, + { + "epoch": 2.482532751091703, + "grad_norm": 0.235432313345855, + "learning_rate": 4.4035027487812325e-06, + "loss": 0.2119, + "step": 5685 + }, + { + "epoch": 2.48471615720524, + "grad_norm": 0.19883691503756268, + "learning_rate": 4.367570662884438e-06, + "loss": 0.1997, + "step": 5690 + }, + { + "epoch": 2.486899563318777, + "grad_norm": 0.19289587740272107, + "learning_rate": 4.331771742386395e-06, + "loss": 0.2096, + "step": 5695 + }, + { + "epoch": 2.489082969432314, + "grad_norm": 0.18568302603914705, + "learning_rate": 4.296106218339757e-06, + "loss": 0.2056, + "step": 5700 + }, + { + "epoch": 2.4912663755458517, + "grad_norm": 0.1886391942932925, + "learning_rate": 4.260574320936203e-06, + "loss": 0.2022, + "step": 5705 + }, + { + "epoch": 2.493449781659389, + "grad_norm": 0.19736454250903648, + "learning_rate": 4.225176279504975e-06, + "loss": 0.2139, + "step": 5710 + }, + { + "epoch": 2.495633187772926, + "grad_norm": 0.20848042828630423, + "learning_rate": 4.1899123225113825e-06, + "loss": 0.2033, + "step": 5715 + }, + { + "epoch": 2.497816593886463, + "grad_norm": 0.18156063895028207, + "learning_rate": 4.154782677555324e-06, + "loss": 0.2055, + "step": 5720 + }, + { + "epoch": 2.5, + "grad_norm": 0.1915665995869679, + "learning_rate": 4.119787571369829e-06, + "loss": 0.2056, + "step": 5725 + }, + { + "epoch": 2.502183406113537, + "grad_norm": 0.20789030655261928, + "learning_rate": 4.084927229819591e-06, + "loss": 0.1969, + "step": 5730 + }, + { + "epoch": 2.504366812227074, + "grad_norm": 0.19567843626163597, + "learning_rate": 4.050201877899501e-06, + "loss": 0.1967, + "step": 5735 + }, + { + "epoch": 2.506550218340611, + "grad_norm": 0.19524615840080548, + "learning_rate": 4.0156117397332074e-06, + "loss": 0.2017, + "step": 5740 + }, + { + "epoch": 2.5087336244541483, + "grad_norm": 0.1851828139613455, + "learning_rate": 3.9811570385716705e-06, + "loss": 0.2187, + "step": 5745 + }, + { + "epoch": 2.5109170305676853, + "grad_norm": 0.18768129563315683, + "learning_rate": 3.9468379967917025e-06, + "loss": 0.2139, + "step": 5750 + }, + { + "epoch": 2.513100436681223, + "grad_norm": 0.1908316174021218, + "learning_rate": 3.912654835894564e-06, + "loss": 0.2085, + "step": 5755 + }, + { + "epoch": 2.51528384279476, + "grad_norm": 0.18887997682497717, + "learning_rate": 3.878607776504489e-06, + "loss": 0.2089, + "step": 5760 + }, + { + "epoch": 2.517467248908297, + "grad_norm": 0.18931507612361673, + "learning_rate": 3.844697038367314e-06, + "loss": 0.2118, + "step": 5765 + }, + { + "epoch": 2.519650655021834, + "grad_norm": 0.1955389549996342, + "learning_rate": 3.8109228403490268e-06, + "loss": 0.2041, + "step": 5770 + }, + { + "epoch": 2.521834061135371, + "grad_norm": 0.1985311672510908, + "learning_rate": 3.7772854004343494e-06, + "loss": 0.2117, + "step": 5775 + }, + { + "epoch": 2.524017467248908, + "grad_norm": 0.18193407723396662, + "learning_rate": 3.7437849357253555e-06, + "loss": 0.2051, + "step": 5780 + }, + { + "epoch": 2.5262008733624453, + "grad_norm": 0.19363438642248365, + "learning_rate": 3.7104216624400502e-06, + "loss": 0.2112, + "step": 5785 + }, + { + "epoch": 2.5283842794759828, + "grad_norm": 0.18728937794992326, + "learning_rate": 3.677195795910984e-06, + "loss": 0.2052, + "step": 5790 + }, + { + "epoch": 2.53056768558952, + "grad_norm": 0.17522592702667172, + "learning_rate": 3.6441075505838586e-06, + "loss": 0.2108, + "step": 5795 + }, + { + "epoch": 2.532751091703057, + "grad_norm": 0.19669895879396115, + "learning_rate": 3.6111571400161387e-06, + "loss": 0.2107, + "step": 5800 + }, + { + "epoch": 2.534934497816594, + "grad_norm": 0.2086966810335374, + "learning_rate": 3.578344776875689e-06, + "loss": 0.2025, + "step": 5805 + }, + { + "epoch": 2.537117903930131, + "grad_norm": 0.18427750014405722, + "learning_rate": 3.5456706729393774e-06, + "loss": 0.2089, + "step": 5810 + }, + { + "epoch": 2.539301310043668, + "grad_norm": 0.19759644279021854, + "learning_rate": 3.5131350390917333e-06, + "loss": 0.2136, + "step": 5815 + }, + { + "epoch": 2.541484716157205, + "grad_norm": 0.18832000886509567, + "learning_rate": 3.4807380853235704e-06, + "loss": 0.2206, + "step": 5820 + }, + { + "epoch": 2.5436681222707422, + "grad_norm": 0.1939256790542837, + "learning_rate": 3.4484800207306405e-06, + "loss": 0.2104, + "step": 5825 + }, + { + "epoch": 2.5458515283842793, + "grad_norm": 0.1981092726688764, + "learning_rate": 3.4163610535122738e-06, + "loss": 0.2082, + "step": 5830 + }, + { + "epoch": 2.5480349344978164, + "grad_norm": 0.20236118691003865, + "learning_rate": 3.384381390970048e-06, + "loss": 0.2046, + "step": 5835 + }, + { + "epoch": 2.5502183406113534, + "grad_norm": 0.19004473471111177, + "learning_rate": 3.3525412395064375e-06, + "loss": 0.2023, + "step": 5840 + }, + { + "epoch": 2.552401746724891, + "grad_norm": 0.20444360368222714, + "learning_rate": 3.32084080462349e-06, + "loss": 0.2027, + "step": 5845 + }, + { + "epoch": 2.554585152838428, + "grad_norm": 0.2175490290108786, + "learning_rate": 3.2892802909214955e-06, + "loss": 0.2096, + "step": 5850 + }, + { + "epoch": 2.556768558951965, + "grad_norm": 0.19247531436865034, + "learning_rate": 3.2578599020976707e-06, + "loss": 0.1951, + "step": 5855 + }, + { + "epoch": 2.558951965065502, + "grad_norm": 0.1874278832882864, + "learning_rate": 3.2265798409448376e-06, + "loss": 0.211, + "step": 5860 + }, + { + "epoch": 2.561135371179039, + "grad_norm": 0.19407368076943168, + "learning_rate": 3.1954403093501227e-06, + "loss": 0.2179, + "step": 5865 + }, + { + "epoch": 2.5633187772925763, + "grad_norm": 0.176804419787979, + "learning_rate": 3.1644415082936417e-06, + "loss": 0.2017, + "step": 5870 + }, + { + "epoch": 2.565502183406114, + "grad_norm": 0.1936425739896731, + "learning_rate": 3.1335836378472233e-06, + "loss": 0.1963, + "step": 5875 + }, + { + "epoch": 2.567685589519651, + "grad_norm": 0.18623478504104157, + "learning_rate": 3.10286689717309e-06, + "loss": 0.2076, + "step": 5880 + }, + { + "epoch": 2.569868995633188, + "grad_norm": 0.18973768721599718, + "learning_rate": 3.0722914845225963e-06, + "loss": 0.211, + "step": 5885 + }, + { + "epoch": 2.572052401746725, + "grad_norm": 0.1887388707833972, + "learning_rate": 3.04185759723494e-06, + "loss": 0.1983, + "step": 5890 + }, + { + "epoch": 2.574235807860262, + "grad_norm": 0.18187528811137918, + "learning_rate": 3.0115654317358884e-06, + "loss": 0.2119, + "step": 5895 + }, + { + "epoch": 2.576419213973799, + "grad_norm": 0.19057590033301217, + "learning_rate": 2.981415183536504e-06, + "loss": 0.212, + "step": 5900 + }, + { + "epoch": 2.578602620087336, + "grad_norm": 0.18728042390500685, + "learning_rate": 2.9514070472318965e-06, + "loss": 0.1956, + "step": 5905 + }, + { + "epoch": 2.5807860262008733, + "grad_norm": 0.18967646290907864, + "learning_rate": 2.9215412164999594e-06, + "loss": 0.2006, + "step": 5910 + }, + { + "epoch": 2.5829694323144103, + "grad_norm": 0.19120785023883197, + "learning_rate": 2.891817884100123e-06, + "loss": 0.192, + "step": 5915 + }, + { + "epoch": 2.5851528384279474, + "grad_norm": 0.18671520762510882, + "learning_rate": 2.862237241872101e-06, + "loss": 0.2215, + "step": 5920 + }, + { + "epoch": 2.5873362445414845, + "grad_norm": 0.1891615762513396, + "learning_rate": 2.832799480734663e-06, + "loss": 0.2244, + "step": 5925 + }, + { + "epoch": 2.589519650655022, + "grad_norm": 0.1807780047483408, + "learning_rate": 2.8035047906844e-06, + "loss": 0.2053, + "step": 5930 + }, + { + "epoch": 2.591703056768559, + "grad_norm": 0.18284608120565451, + "learning_rate": 2.774353360794493e-06, + "loss": 0.2047, + "step": 5935 + }, + { + "epoch": 2.593886462882096, + "grad_norm": 0.18110685745874158, + "learning_rate": 2.7453453792134986e-06, + "loss": 0.213, + "step": 5940 + }, + { + "epoch": 2.596069868995633, + "grad_norm": 0.18936818198939037, + "learning_rate": 2.7164810331641366e-06, + "loss": 0.2038, + "step": 5945 + }, + { + "epoch": 2.5982532751091703, + "grad_norm": 0.19029850620608788, + "learning_rate": 2.687760508942064e-06, + "loss": 0.2116, + "step": 5950 + }, + { + "epoch": 2.6004366812227073, + "grad_norm": 0.19115405588229892, + "learning_rate": 2.659183991914696e-06, + "loss": 0.2145, + "step": 5955 + }, + { + "epoch": 2.6026200873362444, + "grad_norm": 0.1945513238775359, + "learning_rate": 2.6307516665200012e-06, + "loss": 0.208, + "step": 5960 + }, + { + "epoch": 2.604803493449782, + "grad_norm": 0.19282829361579937, + "learning_rate": 2.602463716265302e-06, + "loss": 0.2179, + "step": 5965 + }, + { + "epoch": 2.606986899563319, + "grad_norm": 0.2002570604911309, + "learning_rate": 2.5743203237261065e-06, + "loss": 0.2095, + "step": 5970 + }, + { + "epoch": 2.609170305676856, + "grad_norm": 0.18334332369689957, + "learning_rate": 2.5463216705449155e-06, + "loss": 0.2185, + "step": 5975 + }, + { + "epoch": 2.611353711790393, + "grad_norm": 0.181736048243745, + "learning_rate": 2.5184679374300553e-06, + "loss": 0.2085, + "step": 5980 + }, + { + "epoch": 2.61353711790393, + "grad_norm": 0.17385940038429637, + "learning_rate": 2.4907593041545163e-06, + "loss": 0.2167, + "step": 5985 + }, + { + "epoch": 2.6157205240174672, + "grad_norm": 0.17922911028497665, + "learning_rate": 2.4631959495547833e-06, + "loss": 0.1995, + "step": 5990 + }, + { + "epoch": 2.6179039301310043, + "grad_norm": 0.17779619304763405, + "learning_rate": 2.4357780515297e-06, + "loss": 0.1941, + "step": 5995 + }, + { + "epoch": 2.6200873362445414, + "grad_norm": 0.19092236134949003, + "learning_rate": 2.408505787039281e-06, + "loss": 0.2023, + "step": 6000 + }, + { + "epoch": 2.6222707423580784, + "grad_norm": 0.21537785339457424, + "learning_rate": 2.381379332103623e-06, + "loss": 0.2039, + "step": 6005 + }, + { + "epoch": 2.6244541484716155, + "grad_norm": 0.19076636501936398, + "learning_rate": 2.3543988618017237e-06, + "loss": 0.2007, + "step": 6010 + }, + { + "epoch": 2.6266375545851526, + "grad_norm": 0.18091299115157525, + "learning_rate": 2.327564550270381e-06, + "loss": 0.2003, + "step": 6015 + }, + { + "epoch": 2.62882096069869, + "grad_norm": 0.17652429068296252, + "learning_rate": 2.3008765707030532e-06, + "loss": 0.1987, + "step": 6020 + }, + { + "epoch": 2.631004366812227, + "grad_norm": 0.19337565491737743, + "learning_rate": 2.2743350953487424e-06, + "loss": 0.2161, + "step": 6025 + }, + { + "epoch": 2.6331877729257642, + "grad_norm": 0.19521279815086381, + "learning_rate": 2.2479402955108945e-06, + "loss": 0.1994, + "step": 6030 + }, + { + "epoch": 2.6353711790393013, + "grad_norm": 0.18448803162094932, + "learning_rate": 2.2216923415462802e-06, + "loss": 0.2071, + "step": 6035 + }, + { + "epoch": 2.6375545851528384, + "grad_norm": 0.19952616634096593, + "learning_rate": 2.1955914028639e-06, + "loss": 0.1997, + "step": 6040 + }, + { + "epoch": 2.6397379912663754, + "grad_norm": 0.17820128872734733, + "learning_rate": 2.1696376479238947e-06, + "loss": 0.2095, + "step": 6045 + }, + { + "epoch": 2.641921397379913, + "grad_norm": 0.18942931129920965, + "learning_rate": 2.1438312442364467e-06, + "loss": 0.2065, + "step": 6050 + }, + { + "epoch": 2.64410480349345, + "grad_norm": 0.1880722865584005, + "learning_rate": 2.1181723583607177e-06, + "loss": 0.2011, + "step": 6055 + }, + { + "epoch": 2.646288209606987, + "grad_norm": 0.19525082185206008, + "learning_rate": 2.092661155903758e-06, + "loss": 0.2075, + "step": 6060 + }, + { + "epoch": 2.648471615720524, + "grad_norm": 0.17350084934940418, + "learning_rate": 2.067297801519441e-06, + "loss": 0.1998, + "step": 6065 + }, + { + "epoch": 2.650655021834061, + "grad_norm": 0.18323734039467143, + "learning_rate": 2.0420824589074073e-06, + "loss": 0.2066, + "step": 6070 + }, + { + "epoch": 2.6528384279475983, + "grad_norm": 0.19042032301324274, + "learning_rate": 2.0170152908119964e-06, + "loss": 0.2078, + "step": 6075 + }, + { + "epoch": 2.6550218340611353, + "grad_norm": 0.1822344566119162, + "learning_rate": 1.9920964590212088e-06, + "loss": 0.2026, + "step": 6080 + }, + { + "epoch": 2.6572052401746724, + "grad_norm": 0.18350664419747995, + "learning_rate": 1.967326124365651e-06, + "loss": 0.2046, + "step": 6085 + }, + { + "epoch": 2.6593886462882095, + "grad_norm": 0.17900525181219676, + "learning_rate": 1.9427044467175054e-06, + "loss": 0.2053, + "step": 6090 + }, + { + "epoch": 2.6615720524017465, + "grad_norm": 0.18662259458308622, + "learning_rate": 1.9182315849894996e-06, + "loss": 0.1969, + "step": 6095 + }, + { + "epoch": 2.6637554585152836, + "grad_norm": 0.18210726045913153, + "learning_rate": 1.8939076971338592e-06, + "loss": 0.2133, + "step": 6100 + }, + { + "epoch": 2.665938864628821, + "grad_norm": 0.18936892016307003, + "learning_rate": 1.8697329401413333e-06, + "loss": 0.2036, + "step": 6105 + }, + { + "epoch": 2.668122270742358, + "grad_norm": 0.17873699633744589, + "learning_rate": 1.8457074700401284e-06, + "loss": 0.2015, + "step": 6110 + }, + { + "epoch": 2.6703056768558953, + "grad_norm": 0.19259954300544802, + "learning_rate": 1.821831441894939e-06, + "loss": 0.2168, + "step": 6115 + }, + { + "epoch": 2.6724890829694323, + "grad_norm": 0.1765084388420357, + "learning_rate": 1.7981050098059364e-06, + "loss": 0.2051, + "step": 6120 + }, + { + "epoch": 2.6746724890829694, + "grad_norm": 0.18953853048678723, + "learning_rate": 1.7745283269077622e-06, + "loss": 0.1945, + "step": 6125 + }, + { + "epoch": 2.6768558951965065, + "grad_norm": 0.17824046665997959, + "learning_rate": 1.751101545368558e-06, + "loss": 0.1987, + "step": 6130 + }, + { + "epoch": 2.6790393013100435, + "grad_norm": 0.1839474436743472, + "learning_rate": 1.7278248163889787e-06, + "loss": 0.2297, + "step": 6135 + }, + { + "epoch": 2.681222707423581, + "grad_norm": 0.198016592733994, + "learning_rate": 1.7046982902011954e-06, + "loss": 0.2004, + "step": 6140 + }, + { + "epoch": 2.683406113537118, + "grad_norm": 0.1754478072579575, + "learning_rate": 1.6817221160679603e-06, + "loss": 0.2165, + "step": 6145 + }, + { + "epoch": 2.685589519650655, + "grad_norm": 0.1739282991837851, + "learning_rate": 1.658896442281624e-06, + "loss": 0.2139, + "step": 6150 + }, + { + "epoch": 2.6877729257641922, + "grad_norm": 0.1752625270758878, + "learning_rate": 1.636221416163175e-06, + "loss": 0.1918, + "step": 6155 + }, + { + "epoch": 2.6899563318777293, + "grad_norm": 0.1779820497152234, + "learning_rate": 1.6136971840612996e-06, + "loss": 0.1899, + "step": 6160 + }, + { + "epoch": 2.6921397379912664, + "grad_norm": 0.18605348108430886, + "learning_rate": 1.5913238913514317e-06, + "loss": 0.2034, + "step": 6165 + }, + { + "epoch": 2.6943231441048034, + "grad_norm": 0.18327039260922082, + "learning_rate": 1.569101682434812e-06, + "loss": 0.2151, + "step": 6170 + }, + { + "epoch": 2.6965065502183405, + "grad_norm": 0.18420649704706576, + "learning_rate": 1.5470307007375617e-06, + "loss": 0.1999, + "step": 6175 + }, + { + "epoch": 2.6986899563318776, + "grad_norm": 0.18496496761599715, + "learning_rate": 1.5251110887097547e-06, + "loss": 0.2015, + "step": 6180 + }, + { + "epoch": 2.7008733624454146, + "grad_norm": 0.18359525195308157, + "learning_rate": 1.5033429878245047e-06, + "loss": 0.1964, + "step": 6185 + }, + { + "epoch": 2.7030567685589517, + "grad_norm": 0.18877940096775106, + "learning_rate": 1.4817265385770273e-06, + "loss": 0.219, + "step": 6190 + }, + { + "epoch": 2.7052401746724892, + "grad_norm": 0.17210679893859862, + "learning_rate": 1.460261880483768e-06, + "loss": 0.2087, + "step": 6195 + }, + { + "epoch": 2.7074235807860263, + "grad_norm": 0.18497422573663475, + "learning_rate": 1.4389491520814785e-06, + "loss": 0.2073, + "step": 6200 + }, + { + "epoch": 2.7096069868995634, + "grad_norm": 0.18280651214761798, + "learning_rate": 1.417788490926328e-06, + "loss": 0.2049, + "step": 6205 + }, + { + "epoch": 2.7117903930131004, + "grad_norm": 0.21697399944996257, + "learning_rate": 1.3967800335930203e-06, + "loss": 0.1966, + "step": 6210 + }, + { + "epoch": 2.7139737991266375, + "grad_norm": 0.17484224090531814, + "learning_rate": 1.3759239156739046e-06, + "loss": 0.2004, + "step": 6215 + }, + { + "epoch": 2.7161572052401746, + "grad_norm": 0.17922630540901408, + "learning_rate": 1.355220271778107e-06, + "loss": 0.2171, + "step": 6220 + }, + { + "epoch": 2.7183406113537116, + "grad_norm": 0.1788064963960434, + "learning_rate": 1.3346692355306583e-06, + "loss": 0.21, + "step": 6225 + }, + { + "epoch": 2.720524017467249, + "grad_norm": 0.17658228011851998, + "learning_rate": 1.3142709395716325e-06, + "loss": 0.2041, + "step": 6230 + }, + { + "epoch": 2.722707423580786, + "grad_norm": 0.19747059238001105, + "learning_rate": 1.294025515555289e-06, + "loss": 0.2141, + "step": 6235 + }, + { + "epoch": 2.7248908296943233, + "grad_norm": 0.19234938030154514, + "learning_rate": 1.2739330941492211e-06, + "loss": 0.2194, + "step": 6240 + }, + { + "epoch": 2.7270742358078603, + "grad_norm": 0.18116820313254028, + "learning_rate": 1.253993805033521e-06, + "loss": 0.2026, + "step": 6245 + }, + { + "epoch": 2.7292576419213974, + "grad_norm": 0.18091742916208972, + "learning_rate": 1.2342077768999372e-06, + "loss": 0.2074, + "step": 6250 + }, + { + "epoch": 2.7314410480349345, + "grad_norm": 0.18173772524843324, + "learning_rate": 1.2145751374510462e-06, + "loss": 0.1982, + "step": 6255 + }, + { + "epoch": 2.7336244541484715, + "grad_norm": 0.18111352269398215, + "learning_rate": 1.1950960133994189e-06, + "loss": 0.2113, + "step": 6260 + }, + { + "epoch": 2.7358078602620086, + "grad_norm": 0.17797289847881081, + "learning_rate": 1.175770530466816e-06, + "loss": 0.1985, + "step": 6265 + }, + { + "epoch": 2.7379912663755457, + "grad_norm": 0.18092908617426118, + "learning_rate": 1.156598813383372e-06, + "loss": 0.2228, + "step": 6270 + }, + { + "epoch": 2.7401746724890828, + "grad_norm": 0.18444520409610834, + "learning_rate": 1.1375809858867853e-06, + "loss": 0.2114, + "step": 6275 + }, + { + "epoch": 2.74235807860262, + "grad_norm": 0.1725831612805252, + "learning_rate": 1.1187171707215228e-06, + "loss": 0.2016, + "step": 6280 + }, + { + "epoch": 2.7445414847161573, + "grad_norm": 0.1822891732232196, + "learning_rate": 1.1000074896380386e-06, + "loss": 0.2046, + "step": 6285 + }, + { + "epoch": 2.7467248908296944, + "grad_norm": 0.1837802202840671, + "learning_rate": 1.0814520633919577e-06, + "loss": 0.2063, + "step": 6290 + }, + { + "epoch": 2.7489082969432315, + "grad_norm": 0.19854848233423214, + "learning_rate": 1.063051011743335e-06, + "loss": 0.2026, + "step": 6295 + }, + { + "epoch": 2.7510917030567685, + "grad_norm": 0.1690617798747192, + "learning_rate": 1.0448044534558582e-06, + "loss": 0.2042, + "step": 6300 + }, + { + "epoch": 2.7532751091703056, + "grad_norm": 0.18680449525384113, + "learning_rate": 1.026712506296082e-06, + "loss": 0.2198, + "step": 6305 + }, + { + "epoch": 2.7554585152838427, + "grad_norm": 0.1690572517735115, + "learning_rate": 1.0087752870326817e-06, + "loss": 0.2128, + "step": 6310 + }, + { + "epoch": 2.75764192139738, + "grad_norm": 0.1741781541277763, + "learning_rate": 9.909929114356842e-07, + "loss": 0.2026, + "step": 6315 + }, + { + "epoch": 2.7598253275109172, + "grad_norm": 0.17546914742624392, + "learning_rate": 9.733654942757353e-07, + "loss": 0.1994, + "step": 6320 + }, + { + "epoch": 2.7620087336244543, + "grad_norm": 0.17660338305720158, + "learning_rate": 9.558931493233392e-07, + "loss": 0.204, + "step": 6325 + }, + { + "epoch": 2.7641921397379914, + "grad_norm": 0.1769127719144767, + "learning_rate": 9.385759893481477e-07, + "loss": 0.1996, + "step": 6330 + }, + { + "epoch": 2.7663755458515285, + "grad_norm": 0.17837387586213685, + "learning_rate": 9.214141261182196e-07, + "loss": 0.2156, + "step": 6335 + }, + { + "epoch": 2.7685589519650655, + "grad_norm": 0.18449098828135574, + "learning_rate": 9.044076703992904e-07, + "loss": 0.208, + "step": 6340 + }, + { + "epoch": 2.7707423580786026, + "grad_norm": 0.18352552163108934, + "learning_rate": 8.875567319540784e-07, + "loss": 0.2074, + "step": 6345 + }, + { + "epoch": 2.7729257641921397, + "grad_norm": 0.18503191612346148, + "learning_rate": 8.708614195415604e-07, + "loss": 0.2, + "step": 6350 + }, + { + "epoch": 2.7751091703056767, + "grad_norm": 0.1937080570900333, + "learning_rate": 8.543218409162778e-07, + "loss": 0.2082, + "step": 6355 + }, + { + "epoch": 2.777292576419214, + "grad_norm": 0.18510996106818742, + "learning_rate": 8.379381028276401e-07, + "loss": 0.2056, + "step": 6360 + }, + { + "epoch": 2.779475982532751, + "grad_norm": 0.1777159695537935, + "learning_rate": 8.217103110192276e-07, + "loss": 0.2061, + "step": 6365 + }, + { + "epoch": 2.7816593886462884, + "grad_norm": 0.1752542433085556, + "learning_rate": 8.056385702281177e-07, + "loss": 0.2007, + "step": 6370 + }, + { + "epoch": 2.7838427947598254, + "grad_norm": 0.1789778389216568, + "learning_rate": 7.89722984184213e-07, + "loss": 0.2046, + "step": 6375 + }, + { + "epoch": 2.7860262008733625, + "grad_norm": 0.17685450533969785, + "learning_rate": 7.739636556095525e-07, + "loss": 0.2065, + "step": 6380 + }, + { + "epoch": 2.7882096069868996, + "grad_norm": 0.186062026864878, + "learning_rate": 7.583606862176712e-07, + "loss": 0.2032, + "step": 6385 + }, + { + "epoch": 2.7903930131004366, + "grad_norm": 0.16820621198730876, + "learning_rate": 7.429141767129333e-07, + "loss": 0.2037, + "step": 6390 + }, + { + "epoch": 2.7925764192139737, + "grad_norm": 0.18272612053201714, + "learning_rate": 7.276242267898775e-07, + "loss": 0.2077, + "step": 6395 + }, + { + "epoch": 2.7947598253275108, + "grad_norm": 0.1818746307663275, + "learning_rate": 7.12490935132587e-07, + "loss": 0.2126, + "step": 6400 + }, + { + "epoch": 2.7969432314410483, + "grad_norm": 0.1755002936425013, + "learning_rate": 6.97514399414037e-07, + "loss": 0.2032, + "step": 6405 + }, + { + "epoch": 2.7991266375545854, + "grad_norm": 0.18013946700206887, + "learning_rate": 6.826947162954733e-07, + "loss": 0.2066, + "step": 6410 + }, + { + "epoch": 2.8013100436681224, + "grad_norm": 0.17849038284790902, + "learning_rate": 6.68031981425793e-07, + "loss": 0.2055, + "step": 6415 + }, + { + "epoch": 2.8034934497816595, + "grad_norm": 0.21383468189806515, + "learning_rate": 6.535262894409116e-07, + "loss": 0.2171, + "step": 6420 + }, + { + "epoch": 2.8056768558951966, + "grad_norm": 0.1707908400592636, + "learning_rate": 6.391777339631727e-07, + "loss": 0.1968, + "step": 6425 + }, + { + "epoch": 2.8078602620087336, + "grad_norm": 0.1928726259141998, + "learning_rate": 6.249864076007167e-07, + "loss": 0.2078, + "step": 6430 + }, + { + "epoch": 2.8100436681222707, + "grad_norm": 0.18080723535203885, + "learning_rate": 6.109524019469154e-07, + "loss": 0.2052, + "step": 6435 + }, + { + "epoch": 2.8122270742358078, + "grad_norm": 0.1785838290629133, + "learning_rate": 5.970758075797556e-07, + "loss": 0.1978, + "step": 6440 + }, + { + "epoch": 2.814410480349345, + "grad_norm": 0.17615554784359172, + "learning_rate": 5.833567140612673e-07, + "loss": 0.205, + "step": 6445 + }, + { + "epoch": 2.816593886462882, + "grad_norm": 0.18081617254199714, + "learning_rate": 5.69795209936938e-07, + "loss": 0.2085, + "step": 6450 + }, + { + "epoch": 2.818777292576419, + "grad_norm": 0.19233898776215452, + "learning_rate": 5.563913827351469e-07, + "loss": 0.2094, + "step": 6455 + }, + { + "epoch": 2.8209606986899565, + "grad_norm": 0.18538569170027117, + "learning_rate": 5.431453189666036e-07, + "loss": 0.2133, + "step": 6460 + }, + { + "epoch": 2.8231441048034935, + "grad_norm": 0.16627030997203804, + "learning_rate": 5.30057104123774e-07, + "loss": 0.1946, + "step": 6465 + }, + { + "epoch": 2.8253275109170306, + "grad_norm": 0.18546465605942317, + "learning_rate": 5.171268226803416e-07, + "loss": 0.2014, + "step": 6470 + }, + { + "epoch": 2.8275109170305677, + "grad_norm": 0.18279009388047487, + "learning_rate": 5.043545580906695e-07, + "loss": 0.2228, + "step": 6475 + }, + { + "epoch": 2.8296943231441047, + "grad_norm": 0.17398692192324666, + "learning_rate": 4.917403927892389e-07, + "loss": 0.208, + "step": 6480 + }, + { + "epoch": 2.831877729257642, + "grad_norm": 0.17416683242727254, + "learning_rate": 4.792844081901309e-07, + "loss": 0.1994, + "step": 6485 + }, + { + "epoch": 2.834061135371179, + "grad_norm": 0.177823166130051, + "learning_rate": 4.6698668468651006e-07, + "loss": 0.1939, + "step": 6490 + }, + { + "epoch": 2.8362445414847164, + "grad_norm": 0.19673421756206907, + "learning_rate": 4.548473016500854e-07, + "loss": 0.2161, + "step": 6495 + }, + { + "epoch": 2.8384279475982535, + "grad_norm": 0.1811334337251195, + "learning_rate": 4.4286633743061413e-07, + "loss": 0.2082, + "step": 6500 + }, + { + "epoch": 2.8406113537117905, + "grad_norm": 0.17564798960003072, + "learning_rate": 4.310438693553853e-07, + "loss": 0.2066, + "step": 6505 + }, + { + "epoch": 2.8427947598253276, + "grad_norm": 0.1794074378421751, + "learning_rate": 4.193799737287285e-07, + "loss": 0.2093, + "step": 6510 + }, + { + "epoch": 2.8449781659388647, + "grad_norm": 0.18373063476787263, + "learning_rate": 4.0787472583151964e-07, + "loss": 0.2235, + "step": 6515 + }, + { + "epoch": 2.8471615720524017, + "grad_norm": 0.17915751006104885, + "learning_rate": 3.965281999206899e-07, + "loss": 0.1956, + "step": 6520 + }, + { + "epoch": 2.849344978165939, + "grad_norm": 0.17324382598114016, + "learning_rate": 3.853404692287538e-07, + "loss": 0.2074, + "step": 6525 + }, + { + "epoch": 2.851528384279476, + "grad_norm": 0.18692578826449793, + "learning_rate": 3.743116059633317e-07, + "loss": 0.2174, + "step": 6530 + }, + { + "epoch": 2.853711790393013, + "grad_norm": 0.18933592310694983, + "learning_rate": 3.6344168130668097e-07, + "loss": 0.2058, + "step": 6535 + }, + { + "epoch": 2.85589519650655, + "grad_norm": 0.18309703282903245, + "learning_rate": 3.527307654152462e-07, + "loss": 0.203, + "step": 6540 + }, + { + "epoch": 2.858078602620087, + "grad_norm": 0.17336689816177772, + "learning_rate": 3.421789274192011e-07, + "loss": 0.189, + "step": 6545 + }, + { + "epoch": 2.8602620087336246, + "grad_norm": 0.17273324095781836, + "learning_rate": 3.3178623542199916e-07, + "loss": 0.2009, + "step": 6550 + }, + { + "epoch": 2.8624454148471616, + "grad_norm": 0.18054845843338757, + "learning_rate": 3.2155275649994045e-07, + "loss": 0.2125, + "step": 6555 + }, + { + "epoch": 2.8646288209606987, + "grad_norm": 0.17486255516315685, + "learning_rate": 3.1147855670172755e-07, + "loss": 0.2014, + "step": 6560 + }, + { + "epoch": 2.8668122270742358, + "grad_norm": 0.16994065735909195, + "learning_rate": 3.015637010480576e-07, + "loss": 0.1991, + "step": 6565 + }, + { + "epoch": 2.868995633187773, + "grad_norm": 0.17823602403047226, + "learning_rate": 2.918082535311839e-07, + "loss": 0.2107, + "step": 6570 + }, + { + "epoch": 2.87117903930131, + "grad_norm": 0.18534656992358386, + "learning_rate": 2.822122771145158e-07, + "loss": 0.1988, + "step": 6575 + }, + { + "epoch": 2.8733624454148474, + "grad_norm": 0.17359168867052116, + "learning_rate": 2.727758337322056e-07, + "loss": 0.2006, + "step": 6580 + }, + { + "epoch": 2.8755458515283845, + "grad_norm": 0.1818021640887144, + "learning_rate": 2.6349898428875143e-07, + "loss": 0.2107, + "step": 6585 + }, + { + "epoch": 2.8777292576419216, + "grad_norm": 0.187022315664445, + "learning_rate": 2.543817886586003e-07, + "loss": 0.2109, + "step": 6590 + }, + { + "epoch": 2.8799126637554586, + "grad_norm": 0.18065592152437351, + "learning_rate": 2.454243056857708e-07, + "loss": 0.2155, + "step": 6595 + }, + { + "epoch": 2.8820960698689957, + "grad_norm": 0.1803124984333343, + "learning_rate": 2.366265931834588e-07, + "loss": 0.2005, + "step": 6600 + }, + { + "epoch": 2.8842794759825328, + "grad_norm": 0.17978461103858198, + "learning_rate": 2.2798870793368787e-07, + "loss": 0.2135, + "step": 6605 + }, + { + "epoch": 2.88646288209607, + "grad_norm": 0.1838336256596766, + "learning_rate": 2.1951070568691222e-07, + "loss": 0.2131, + "step": 6610 + }, + { + "epoch": 2.888646288209607, + "grad_norm": 0.18102300117813305, + "learning_rate": 2.1119264116168104e-07, + "loss": 0.2064, + "step": 6615 + }, + { + "epoch": 2.890829694323144, + "grad_norm": 0.1743140768844891, + "learning_rate": 2.030345680442719e-07, + "loss": 0.1962, + "step": 6620 + }, + { + "epoch": 2.893013100436681, + "grad_norm": 0.18304778762459922, + "learning_rate": 1.950365389883524e-07, + "loss": 0.2145, + "step": 6625 + }, + { + "epoch": 2.895196506550218, + "grad_norm": 0.17457199470182785, + "learning_rate": 1.8719860561463853e-07, + "loss": 0.1986, + "step": 6630 + }, + { + "epoch": 2.8973799126637556, + "grad_norm": 0.17803797345436936, + "learning_rate": 1.7952081851055346e-07, + "loss": 0.2087, + "step": 6635 + }, + { + "epoch": 2.8995633187772927, + "grad_norm": 0.19249639309638913, + "learning_rate": 1.720032272299138e-07, + "loss": 0.208, + "step": 6640 + }, + { + "epoch": 2.9017467248908297, + "grad_norm": 0.1931146601533778, + "learning_rate": 1.6464588029260485e-07, + "loss": 0.2159, + "step": 6645 + }, + { + "epoch": 2.903930131004367, + "grad_norm": 0.17050733886371794, + "learning_rate": 1.5744882518425875e-07, + "loss": 0.2079, + "step": 6650 + }, + { + "epoch": 2.906113537117904, + "grad_norm": 0.1821050184025676, + "learning_rate": 1.5041210835596287e-07, + "loss": 0.2113, + "step": 6655 + }, + { + "epoch": 2.908296943231441, + "grad_norm": 0.1767629872679953, + "learning_rate": 1.4353577522394635e-07, + "loss": 0.2043, + "step": 6660 + }, + { + "epoch": 2.910480349344978, + "grad_norm": 0.1770612573000096, + "learning_rate": 1.3681987016929963e-07, + "loss": 0.2086, + "step": 6665 + }, + { + "epoch": 2.9126637554585155, + "grad_norm": 0.17785323864755895, + "learning_rate": 1.3026443653767762e-07, + "loss": 0.2004, + "step": 6670 + }, + { + "epoch": 2.9148471615720526, + "grad_norm": 0.17642382590921754, + "learning_rate": 1.2386951663901913e-07, + "loss": 0.206, + "step": 6675 + }, + { + "epoch": 2.9170305676855897, + "grad_norm": 0.17596251154223802, + "learning_rate": 1.1763515174729179e-07, + "loss": 0.1986, + "step": 6680 + }, + { + "epoch": 2.9192139737991267, + "grad_norm": 0.1727279406074698, + "learning_rate": 1.115613821002004e-07, + "loss": 0.2038, + "step": 6685 + }, + { + "epoch": 2.921397379912664, + "grad_norm": 0.1761644833661898, + "learning_rate": 1.0564824689894282e-07, + "loss": 0.2117, + "step": 6690 + }, + { + "epoch": 2.923580786026201, + "grad_norm": 0.1772235899771806, + "learning_rate": 9.989578430795732e-08, + "loss": 0.2014, + "step": 6695 + }, + { + "epoch": 2.925764192139738, + "grad_norm": 0.17942614591659065, + "learning_rate": 9.430403145466448e-08, + "loss": 0.2109, + "step": 6700 + }, + { + "epoch": 2.927947598253275, + "grad_norm": 0.17253943064578878, + "learning_rate": 8.887302442924517e-08, + "loss": 0.2003, + "step": 6705 + }, + { + "epoch": 2.930131004366812, + "grad_norm": 0.17852850650326985, + "learning_rate": 8.360279828438789e-08, + "loss": 0.2094, + "step": 6710 + }, + { + "epoch": 2.932314410480349, + "grad_norm": 0.17307368947339777, + "learning_rate": 7.849338703508069e-08, + "loss": 0.2064, + "step": 6715 + }, + { + "epoch": 2.934497816593886, + "grad_norm": 0.1793388736193528, + "learning_rate": 7.3544823658378e-08, + "loss": 0.1927, + "step": 6720 + }, + { + "epoch": 2.9366812227074237, + "grad_norm": 0.1803659744144063, + "learning_rate": 6.875714009319523e-08, + "loss": 0.2076, + "step": 6725 + }, + { + "epoch": 2.938864628820961, + "grad_norm": 0.17865387702225788, + "learning_rate": 6.413036724010058e-08, + "loss": 0.2003, + "step": 6730 + }, + { + "epoch": 2.941048034934498, + "grad_norm": 0.17741136523028206, + "learning_rate": 5.966453496111524e-08, + "loss": 0.2144, + "step": 6735 + }, + { + "epoch": 2.943231441048035, + "grad_norm": 0.17550555860513464, + "learning_rate": 5.5359672079524637e-08, + "loss": 0.2153, + "step": 6740 + }, + { + "epoch": 2.945414847161572, + "grad_norm": 0.17696250741970354, + "learning_rate": 5.121580637968137e-08, + "loss": 0.2054, + "step": 6745 + }, + { + "epoch": 2.947598253275109, + "grad_norm": 0.17454155560080192, + "learning_rate": 4.7232964606841454e-08, + "loss": 0.1971, + "step": 6750 + }, + { + "epoch": 2.9497816593886466, + "grad_norm": 0.17698232159770602, + "learning_rate": 4.341117246697834e-08, + "loss": 0.2085, + "step": 6755 + }, + { + "epoch": 2.9519650655021836, + "grad_norm": 0.17354434091915077, + "learning_rate": 3.9750454626624745e-08, + "loss": 0.2137, + "step": 6760 + }, + { + "epoch": 2.9541484716157207, + "grad_norm": 0.17097280016052646, + "learning_rate": 3.6250834712708846e-08, + "loss": 0.1945, + "step": 6765 + }, + { + "epoch": 2.9563318777292578, + "grad_norm": 0.17551121068500722, + "learning_rate": 3.2912335312407226e-08, + "loss": 0.1865, + "step": 6770 + }, + { + "epoch": 2.958515283842795, + "grad_norm": 0.17583508061228037, + "learning_rate": 2.9734977972989407e-08, + "loss": 0.2107, + "step": 6775 + }, + { + "epoch": 2.960698689956332, + "grad_norm": 0.17683109325722057, + "learning_rate": 2.6718783201692964e-08, + "loss": 0.2127, + "step": 6780 + }, + { + "epoch": 2.962882096069869, + "grad_norm": 0.18803887975352115, + "learning_rate": 2.3863770465576418e-08, + "loss": 0.2156, + "step": 6785 + }, + { + "epoch": 2.965065502183406, + "grad_norm": 0.1771870084804161, + "learning_rate": 2.116995819140821e-08, + "loss": 0.2182, + "step": 6790 + }, + { + "epoch": 2.967248908296943, + "grad_norm": 0.18459436255752454, + "learning_rate": 1.8637363765536264e-08, + "loss": 0.2009, + "step": 6795 + }, + { + "epoch": 2.96943231441048, + "grad_norm": 0.17264295233596325, + "learning_rate": 1.6266003533774166e-08, + "loss": 0.201, + "step": 6800 + }, + { + "epoch": 2.9716157205240172, + "grad_norm": 0.18069190701339236, + "learning_rate": 1.405589280130959e-08, + "loss": 0.214, + "step": 6805 + }, + { + "epoch": 2.9737991266375547, + "grad_norm": 0.17111956645749835, + "learning_rate": 1.2007045832590492e-08, + "loss": 0.1958, + "step": 6810 + }, + { + "epoch": 2.975982532751092, + "grad_norm": 0.17292900306530656, + "learning_rate": 1.0119475851247395e-08, + "loss": 0.1952, + "step": 6815 + }, + { + "epoch": 2.978165938864629, + "grad_norm": 0.1806661006510654, + "learning_rate": 8.393195039993473e-09, + "loss": 0.207, + "step": 6820 + }, + { + "epoch": 2.980349344978166, + "grad_norm": 0.1805502156590632, + "learning_rate": 6.82821454055238e-09, + "loss": 0.2077, + "step": 6825 + }, + { + "epoch": 2.982532751091703, + "grad_norm": 0.17317279258008123, + "learning_rate": 5.42454445359164e-09, + "loss": 0.2, + "step": 6830 + }, + { + "epoch": 2.98471615720524, + "grad_norm": 0.17476834735543392, + "learning_rate": 4.1821938386477075e-09, + "loss": 0.2028, + "step": 6835 + }, + { + "epoch": 2.986899563318777, + "grad_norm": 0.18884495387086833, + "learning_rate": 3.1011707140732317e-09, + "loss": 0.2119, + "step": 6840 + }, + { + "epoch": 2.9890829694323147, + "grad_norm": 0.17552656950302298, + "learning_rate": 2.1814820569898698e-09, + "loss": 0.2146, + "step": 6845 + }, + { + "epoch": 2.9912663755458517, + "grad_norm": 0.17245278444580078, + "learning_rate": 1.4231338032300034e-09, + "loss": 0.213, + "step": 6850 + }, + { + "epoch": 2.993449781659389, + "grad_norm": 0.17656984635938758, + "learning_rate": 8.261308473089813e-10, + "loss": 0.2045, + "step": 6855 + }, + { + "epoch": 2.995633187772926, + "grad_norm": 0.18334478736300916, + "learning_rate": 3.9047704239181247e-10, + "loss": 0.2067, + "step": 6860 + }, + { + "epoch": 2.997816593886463, + "grad_norm": 0.1826422374600176, + "learning_rate": 1.1617520026818706e-10, + "loss": 0.1943, + "step": 6865 + }, + { + "epoch": 3.0, + "grad_norm": 0.1888467433612867, + "learning_rate": 3.2270913274956573e-12, + "loss": 0.1971, + "step": 6870 + }, + { + "epoch": 3.0, + "step": 6870, + "total_flos": 1.440651216420864e+16, + "train_loss": 0.30719185047274594, + "train_runtime": 200812.4774, + "train_samples_per_second": 1.094, + "train_steps_per_second": 0.034 + } + ], + "logging_steps": 5, + "max_steps": 6870, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 10000000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 1.440651216420864e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000..5f17a22 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19ede9ec5d95c9ba6d1490e06a35ab83d954077f265d7ad42d2b81375c33da7b +size 7736 diff --git a/training_loss.png b/training_loss.png new file mode 100644 index 0000000..3109cf6 Binary files /dev/null and b/training_loss.png differ diff --git a/vocab.json b/vocab.json new file mode 100644 index 0000000..6c49fc6 --- /dev/null +++ b/vocab.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910 +size 2776833