Nirbhayhero07/deepsentinel-overseer-small
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
main
deepsentinel-overseer-small/artifacts/train.log
ModelHub XC 42a8f75134 初始化项目，由ModelHub XC社区提供模型 
Model: Nirbhayhero07/deepsentinel-overseer-small
Source: Original Platform
2026-09-06 18:29:15 +08:00

78 lines
118 KiB
Plaintext
Raw Permalink Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
`torch_dtype` is deprecated! Use `dtype` instead!
============================================================
DeepSentinel GRPO Training
============================================================
[1] Collecting 120 training episodes...
Collected 120 episodes
Distribution: 40 easy, 40 medium, 40 hard
[2] Evaluating heuristic baseline...
easy: avg_reward=-0.3681
medium: avg_reward=0.0049
hard: avg_reward=0.0833
[3] Loading model: Qwen/Qwen2.5-0.5B-Instruct
Loading weights: 0%| | 0/290 [00:00<?, ?it/s]
Loading weights: 0%| | 1/290 [00:00<00:00, 8738.13it/s, Materializing param=model.embed_tokens.weight]
Loading weights: 0%| | 1/290 [00:00<00:00, 2663.05it/s, Materializing param=model.embed_tokens.weight]
Loading weights: 1%| | 2/290 [00:00<00:05, 49.37it/s, Materializing param=model.layers.0.input_layernorm.weight]
Loading weights: 1%| | 2/290 [00:00<00:05, 49.13it/s, Materializing param=model.layers.0.input_layernorm.weight]
Loading weights: 1%| | 3/290 [00:00<00:03, 73.10it/s, Materializing param=model.layers.0.mlp.down_proj.weight]
Loading weights: 1%| | 3/290 [00:00<00:03, 72.82it/s, Materializing param=model.layers.0.mlp.down_proj.weight]
Loading weights: 1%|▏ | 4/290 [00:00<00:03, 94.80it/s, Materializing param=model.layers.0.mlp.gate_proj.weight]
Loading weights: 1%|▏ | 4/290 [00:00<00:03, 94.30it/s, Materializing param=model.layers.0.mlp.gate_proj.weight]
Loading weights: 2%|▏ | 5/290 [00:00<00:02, 109.99it/s, Materializing param=model.layers.0.mlp.up_proj.weight]
Loading weights: 2%|▏ | 5/290 [00:00<00:02, 109.59it/s, Materializing param=model.layers.0.mlp.up_proj.weight]
Loading weights: 2%|▏ | 6/290 [00:00<00:02, 127.87it/s, Materializing param=model.layers.0.post_attention_layernorm.weight]
Loading weights: 2%|▏ | 6/290 [00:00<00:02, 127.48it/s, Materializing param=model.layers.0.post_attention_layernorm.weight]
Loading weights: 2%|▏ | 7/290 [00:00<00:01, 147.08it/s, Materializing param=model.layers.0.self_attn.k_proj.bias]
Loading weights: 2%|▏ | 7/290 [00:00<00:01, 146.62it/s, Materializing param=model.layers.0.self_attn.k_proj.bias]
Loading weights: 3%|▎ | 8/290 [00:00<00:01, 166.08it/s, Materializing param=model.layers.0.self_attn.k_proj.weight]
Loading weights: 3%|▎ | 8/290 [00:00<00:01, 165.40it/s, Materializing param=model.layers.0.self_attn.k_proj.weight]
Loading weights: 3%|▎ | 9/290 [00:00<00:01, 181.47it/s, Materializing param=model.layers.0.self_attn.o_proj.weight]
Loading weights: 3%|▎ | 9/290 [00:00<00:01, 180.92it/s, Materializing param=model.layers.0.self_attn.o_proj.weight]
Loading weights: 3%|▎ | 10/290 [00:00<00:01, 199.46it/s, Materializing param=model.layers.0.self_attn.q_proj.bias]
Loading weights: 3%|▎ | 10/290 [00:00<00:01, 198.86it/s, Materializing param=model.layers.0.self_attn.q_proj.bias]
Loading weights: 4%|▍ | 11/290 [00:00<00:01, 217.67it/s, Materializing param=model.layers.0.self_attn.q_proj.weight]
Loading weights: 4%|▍ | 11/290 [00:00<00:01, 217.09it/s, Materializing param=model.layers.0.self_attn.q_proj.weight]
Loading weights: 4%|▍ | 12/290 [00:00<00:01, 235.72it/s, Materializing param=model.layers.0.self_attn.v_proj.bias]
Loading weights: 4%|▍ | 12/290 [00:00<00:01, 235.11it/s, Materializing param=model.layers.0.self_attn.v_proj.bias]
Loading weights: 4%|▍ | 13/290 [00:00<00:01, 252.59it/s, Materializing param=model.layers.0.self_attn.v_proj.weight]
Loading weights: 4%|▍ | 13/290 [00:00<00:01, 251.63it/s, Materializing param=model.layers.0.self_attn.v_proj.weight]
Loading weights: 5%|▍ | 14/290 [00:00<00:01, 267.61it/s, Materializing param=model.layers.1.input_layernorm.weight]
Loading weights: 5%|▍ | 14/290 [00:00<00:01, 266.62it/s, Materializing param=model.layers.1.input_layernorm.weight]
Loading weights: 5%|▌ | 15/290 [00:00<00:00, 284.00it/s, Materializing param=model.layers.1.mlp.down_proj.weight]
Loading weights: 5%|▌ | 15/290 [00:00<00:00, 283.19it/s, Materializing param=model.layers.1.mlp.down_proj.weight]
Loading weights: 6%|▌ | 16/290 [00:00<00:00, 278.33it/s, Materializing param=model.layers.1.mlp.gate_proj.weight]
Loading weights: 6%|▌ | 16/290 [00:00<00:00, 277.46it/s, Materializing param=model.layers.1.mlp.gate_proj.weight]
Loading weights: 6%|▌ | 17/290 [00:00<00:00, 293.52it/s, Materializing param=model.layers.1.mlp.up_proj.weight]
Loading weights: 6%|▌ | 17/290 [00:00<00:00, 292.79it/s, Materializing param=model.layers.1.mlp.up_proj.weight]
Loading weights: 6%|▌ | 18/290 [00:00<00:00, 297.91it/s, Materializing param=model.layers.1.post_attention_layernorm.weight]
Loading weights: 6%|▌ | 18/290 [00:00<00:00, 297.09it/s, Materializing param=model.layers.1.post_attention_layernorm.weight]
Loading weights: 7%|▋ | 19/290 [00:00<00:00, 312.38it/s, Materializing param=model.layers.1.self_attn.k_proj.bias]
Loading weights: 7%|▋ | 19/290 [00:00<00:00, 311.67it/s, Materializing param=model.layers.1.self_attn.k_proj.bias]
Loading weights: 7%|▋ | 20/290 [00:00<00:00, 326.87it/s, Materializing param=model.layers.1.self_attn.k_proj.weight]
Loading weights: 7%|▋ | 20/290 [00:00<00:00, 325.47it/s, Materializing param=model.layers.1.self_attn.k_proj.weight]
Loading weights: 7%|▋ | 21/290 [00:00<00:00, 340.39it/s, Materializing param=model.layers.1.self_attn.o_proj.weight]
Loading weights: 7%|▋ | 21/290 [00:00<00:00, 339.66it/s, Materializing param=model.layers.1.self_attn.o_proj.weight]
Loading weights: 8%|▊ | 22/290 [00:00<00:00, 354.51it/s, Materializing param=model.layers.1.self_attn.q_proj.bias]
Loading weights: 8%|▊ | 22/290 [00:00<00:00, 353.76it/s, Materializing param=model.layers.1.self_attn.q_proj.bias]
Loading weights: 8%|▊ | 23/290 [00:00<00:00, 367.46it/s, Materializing param=model.layers.1.self_attn.q_proj.weight]
Loading weights: 8%|▊ | 23/290 [00:00<00:00, 366.64it/s, Materializing param=model.layers.1.self_attn.q_proj.weight]
Loading weights: 8%|▊ | 24/290 [00:00<00:00, 376.89it/s, Materializing param=model.layers.1.self_attn.v_proj.bias]
Loading weights: 8%|▊ | 24/290 [00:00<00:00, 375.97it/s, Materializing param=model.layers.1.self_attn.v_proj.bias]
Loading weights: 9%|▊ | 25/290 [00:00<00:00, 388.62it/s, Materializing param=model.layers.1.self_attn.v_proj.weight]
Loading weights: 9%|▊ | 25/290 [00:00<00:00, 387.77it/s, Materializing param=model.layers.1.self_attn.v_proj.weight]
Loading weights: 9%|▉ | 26/290 [00:00<00:00, 399.20it/s, Materializing param=model.layers.2.input_layernorm.weight]
Loading weights: 9%|▉ | 26/290 [00:00<00:00, 397.97it/s, Materializing param=model.layers.2.input_layernorm.weight]
Loading weights: 9%|▉ | 27/290 [00:00<00:00, 410.30it/s, Materializing param=model.layers.2.mlp.down_proj.weight]
Loading weights: 9%|▉ | 27/290 [00:00<00:00, 409.26it/s, Materializing param=model.layers.2.mlp.down_proj.weight]
Loading weights: 10%|▉ | 28/290 [00:00<00:00, 409.22it/s, Materializing param=model.layers.2.mlp.gate_proj.weight]
Loading weights: 10%|▉ | 28/290 [00:00<00:00, 408.18it/s, Materializing param=model.layers.2.mlp.gate_proj.weight]
Loading weights: 10%|█ | 29/290 [00:00<00:00, 411.91it/s, Materializing param=model.layers.2.mlp.up_proj.weight]
Loading weights: 10%|█ | 29/290 [00:00<00:00, 410.73it/s, Materializing param=model.layers.2.mlp.up_proj.weight]
Loading weights: 10%|█ | 30/290 [00:00<00:00, 412.98it/s, Materializing param=model.layers.2.post_attention_layernorm.weight]
Loading weights: 10%|█ | 30/290 [00:00<00:00, 412.20it/s, Materializing param=model.layers.2.post_attention_layernorm.weight]
Loading weights: 11%|█ | 31/290 [00:00<00:00, 423.71it/s, Materializing param=model.layers.2.self_attn.k_proj.bias]
Loading weights: 11%|█ | 31/290 [00:00<00:00, 422.31it/s, Materializing param=model.layers.2.self_attn.k_proj.bias]
Loading weights: 11%|█ | 32/290 [00:00<00:00, 428.15it/s, Materializing param=model.layers.2.self_attn.k_proj.weight]
Loading weights: 11%|█ | 32/290 [00:00<00:00, 424.83it/s, Materializing param=model.layers.2.self_attn.k_proj.weight]
Loading weights: 11%|█▏ | 33/290 [00:00<00:00, 436.10it/s, Materializing param=model.layers.2.self_attn.o_proj.weight]
Loading weights: 11%|█▏ | 33/290 [00:00<00:00, 434.60it/s, Materializing param=model.layers.2.self_attn.o_proj.weight]
Loading weights: 12%|█▏ | 34/290 [00:00<00:00, 444.84it/s, Materializing param=model.layers.2.self_attn.q_proj.bias]
Loading weights: 12%|█▏ | 34/290 [00:00<00:00, 443.73it/s, Materializing param=model.layers.2.self_attn.q_proj.bias]
Loading weights: 12%|█▏ | 35/290 [00:00<00:00, 454.19it/s, Materializing param=model.layers.2.self_attn.q_proj.weight]
Loading weights: 12%|█▏ | 35/290 [00:00<00:00, 453.29it/s, Materializing param=model.layers.2.self_attn.q_proj.weight]
Loading weights: 12%|█▏ | 36/290 [00:00<00:00, 463.36it/s, Materializing param=model.layers.2.self_attn.v_proj.bias]
Loading weights: 12%|█▏ | 36/290 [00:00<00:00, 462.04it/s, Materializing param=model.layers.2.self_attn.v_proj.bias]
Loading weights: 13%|█▎ | 37/290 [00:00<00:00, 473.26it/s, Materializing param=model.layers.2.self_attn.v_proj.weight]
Loading weights: 13%|█▎ | 37/290 [00:00<00:00, 472.24it/s, Materializing param=model.layers.2.self_attn.v_proj.weight]
Loading weights: 13%|█▎ | 38/290 [00:00<00:00, 483.38it/s, Materializing param=model.layers.3.input_layernorm.weight]
Loading weights: 13%|█▎ | 38/290 [00:00<00:00, 482.47it/s, Materializing param=model.layers.3.input_layernorm.weight]
Loading weights: 13%|█▎ | 39/290 [00:00<00:00, 485.34it/s, Materializing param=model.layers.3.mlp.down_proj.weight]
Loading weights: 13%|█▎ | 39/290 [00:00<00:00, 484.39it/s, Materializing param=model.layers.3.mlp.down_proj.weight]
Loading weights: 14%|█▍ | 40/290 [00:00<00:00, 494.19it/s, Materializing param=model.layers.3.mlp.gate_proj.weight]
Loading weights: 14%|█▍ | 40/290 [00:00<00:00, 493.10it/s, Materializing param=model.layers.3.mlp.gate_proj.weight]
Loading weights: 14%|█▍ | 41/290 [00:00<00:00, 485.68it/s, Materializing param=model.layers.3.mlp.up_proj.weight]
Loading weights: 14%|█▍ | 41/290 [00:00<00:00, 484.55it/s, Materializing param=model.layers.3.mlp.up_proj.weight]
Loading weights: 14%|█▍ | 42/290 [00:00<00:00, 490.08it/s, Materializing param=model.layers.3.post_attention_layernorm.weight]
Loading weights: 14%|█▍ | 42/290 [00:00<00:00, 488.88it/s, Materializing param=model.layers.3.post_attention_layernorm.weight]
Loading weights: 15%|█▍ | 43/290 [00:00<00:00, 497.47it/s, Materializing param=model.layers.3.self_attn.k_proj.bias]
Loading weights: 15%|█▍ | 43/290 [00:00<00:00, 494.23it/s, Materializing param=model.layers.3.self_attn.k_proj.bias]
Loading weights: 15%|█▌ | 44/290 [00:00<00:00, 503.94it/s, Materializing param=model.layers.3.self_attn.k_proj.weight]
Loading weights: 15%|█▌ | 44/290 [00:00<00:00, 503.13it/s, Materializing param=model.layers.3.self_attn.k_proj.weight]
Loading weights: 16%|█▌ | 45/290 [00:00<00:00, 513.33it/s, Materializing param=model.layers.3.self_attn.o_proj.weight]
Loading weights: 16%|█▌ | 45/290 [00:00<00:00, 512.60it/s, Materializing param=model.layers.3.self_attn.o_proj.weight]
Loading weights: 16%|█▌ | 46/290 [00:00<00:00, 520.87it/s, Materializing param=model.layers.3.self_attn.q_proj.bias]
Loading weights: 16%|█▌ | 46/290 [00:00<00:00, 519.43it/s, Materializing param=model.layers.3.self_attn.q_proj.bias]
Loading weights: 16%|█▌ | 47/290 [00:00<00:00, 528.81it/s, Materializing param=model.layers.3.self_attn.q_proj.weight]
Loading weights: 16%|█▌ | 47/290 [00:00<00:00, 526.61it/s, Materializing param=model.layers.3.self_attn.q_proj.weight]
Loading weights: 17%|█▋ | 48/290 [00:00<00:00, 535.47it/s, Materializing param=model.layers.3.self_attn.v_proj.bias]
Loading weights: 17%|█▋ | 48/290 [00:00<00:00, 534.52it/s, Materializing param=model.layers.3.self_attn.v_proj.bias]
Loading weights: 17%|█▋ | 49/290 [00:00<00:00, 544.16it/s, Materializing param=model.layers.3.self_attn.v_proj.weight]
Loading weights: 17%|█▋ | 49/290 [00:00<00:00, 543.15it/s, Materializing param=model.layers.3.self_attn.v_proj.weight]
Loading weights: 17%|█▋ | 50/290 [00:00<00:00, 550.33it/s, Materializing param=model.layers.4.input_layernorm.weight]
Loading weights: 17%|█▋ | 50/290 [00:00<00:00, 549.17it/s, Materializing param=model.layers.4.input_layernorm.weight]
Loading weights: 18%|█▊ | 51/290 [00:00<00:00, 557.56it/s, Materializing param=model.layers.4.mlp.down_proj.weight]
Loading weights: 18%|█▊ | 51/290 [00:00<00:00, 556.54it/s, Materializing param=model.layers.4.mlp.down_proj.weight]
Loading weights: 18%|█▊ | 52/290 [00:00<00:00, 537.63it/s, Materializing param=model.layers.4.mlp.gate_proj.weight]
Loading weights: 18%|█▊ | 52/290 [00:00<00:00, 536.42it/s, Materializing param=model.layers.4.mlp.gate_proj.weight]
Loading weights: 18%|█▊ | 53/290 [00:00<00:00, 544.95it/s, Materializing param=model.layers.4.mlp.up_proj.weight]
Loading weights: 18%|█▊ | 53/290 [00:00<00:00, 544.09it/s, Materializing param=model.layers.4.mlp.up_proj.weight]
Loading weights: 19%|█▊ | 54/290 [00:00<00:00, 545.27it/s, Materializing param=model.layers.4.post_attention_layernorm.weight]
Loading weights: 19%|█▊ | 54/290 [00:00<00:00, 543.64it/s, Materializing param=model.layers.4.post_attention_layernorm.weight]
Loading weights: 19%|█▉ | 55/290 [00:00<00:00, 551.87it/s, Materializing param=model.layers.4.self_attn.k_proj.bias]
Loading weights: 19%|█▉ | 55/290 [00:00<00:00, 550.76it/s, Materializing param=model.layers.4.self_attn.k_proj.bias]
Loading weights: 19%|█▉ | 56/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.k_proj.bias]
Loading weights: 19%|█▉ | 56/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.k_proj.weight]
Loading weights: 19%|█▉ | 56/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.k_proj.weight]
Loading weights: 20%|█▉ | 57/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.o_proj.weight]
Loading weights: 20%|█▉ | 57/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.o_proj.weight]
Loading weights: 20%|██ | 58/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.q_proj.bias]
Loading weights: 20%|██ | 58/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.q_proj.bias]
Loading weights: 20%|██ | 59/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.q_proj.weight]
Loading weights: 20%|██ | 59/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.q_proj.weight]
Loading weights: 21%|██ | 60/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.v_proj.bias]
Loading weights: 21%|██ | 60/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.v_proj.bias]
Loading weights: 21%|██ | 61/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.v_proj.weight]
Loading weights: 21%|██ | 61/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.4.self_attn.v_proj.weight]
Loading weights: 21%|██▏ | 62/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.input_layernorm.weight]
Loading weights: 21%|██▏ | 62/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.input_layernorm.weight]
Loading weights: 22%|██▏ | 63/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.down_proj.weight]
Loading weights: 22%|██▏ | 63/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.down_proj.weight]
Loading weights: 22%|██▏ | 64/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.gate_proj.weight]
Loading weights: 22%|██▏ | 64/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.gate_proj.weight]
Loading weights: 22%|██▏ | 65/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.up_proj.weight]
Loading weights: 22%|██▏ | 65/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.mlp.up_proj.weight]
Loading weights: 23%|██▎ | 66/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.post_attention_layernorm.weight]
Loading weights: 23%|██▎ | 66/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.post_attention_layernorm.weight]
Loading weights: 23%|██▎ | 67/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.k_proj.bias]
Loading weights: 23%|██▎ | 67/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.k_proj.bias]
Loading weights: 23%|██▎ | 68/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.k_proj.weight]
Loading weights: 23%|██▎ | 68/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.k_proj.weight]
Loading weights: 24%|██▍ | 69/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.o_proj.weight]
Loading weights: 24%|██▍ | 69/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.o_proj.weight]
Loading weights: 24%|██▍ | 70/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.q_proj.bias]
Loading weights: 24%|██▍ | 70/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.q_proj.bias]
Loading weights: 24%|██▍ | 71/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.q_proj.weight]
Loading weights: 24%|██▍ | 71/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.q_proj.weight]
Loading weights: 25%|██▍ | 72/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.v_proj.bias]
Loading weights: 25%|██▍ | 72/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.v_proj.bias]
Loading weights: 25%|██▌ | 73/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.v_proj.weight]
Loading weights: 25%|██▌ | 73/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.5.self_attn.v_proj.weight]
Loading weights: 26%|██▌ | 74/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.input_layernorm.weight]
Loading weights: 26%|██▌ | 74/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.input_layernorm.weight]
Loading weights: 26%|██▌ | 75/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.down_proj.weight]
Loading weights: 26%|██▌ | 75/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.down_proj.weight]
Loading weights: 26%|██▌ | 76/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.gate_proj.weight]
Loading weights: 26%|██▌ | 76/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.gate_proj.weight]
Loading weights: 27%|██▋ | 77/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.up_proj.weight]
Loading weights: 27%|██▋ | 77/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.mlp.up_proj.weight]
Loading weights: 27%|██▋ | 78/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.post_attention_layernorm.weight]
Loading weights: 27%|██▋ | 78/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.post_attention_layernorm.weight]
Loading weights: 27%|██▋ | 79/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.k_proj.bias]
Loading weights: 27%|██▋ | 79/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.k_proj.bias]
Loading weights: 28%|██▊ | 80/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.k_proj.weight]
Loading weights: 28%|██▊ | 80/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.k_proj.weight]
Loading weights: 28%|██▊ | 81/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.o_proj.weight]
Loading weights: 28%|██▊ | 81/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.o_proj.weight]
Loading weights: 28%|██▊ | 82/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.q_proj.bias]
Loading weights: 28%|██▊ | 82/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.q_proj.bias]
Loading weights: 29%|██▊ | 83/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.q_proj.weight]
Loading weights: 29%|██▊ | 83/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.q_proj.weight]
Loading weights: 29%|██▉ | 84/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.v_proj.bias]
Loading weights: 29%|██▉ | 84/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.v_proj.bias]
Loading weights: 29%|██▉ | 85/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.v_proj.weight]
Loading weights: 29%|██▉ | 85/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.6.self_attn.v_proj.weight]
Loading weights: 30%|██▉ | 86/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.input_layernorm.weight]
Loading weights: 30%|██▉ | 86/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.input_layernorm.weight]
Loading weights: 30%|███ | 87/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.down_proj.weight]
Loading weights: 30%|███ | 87/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.down_proj.weight]
Loading weights: 30%|███ | 88/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.gate_proj.weight]
Loading weights: 30%|███ | 88/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.gate_proj.weight]
Loading weights: 31%|███ | 89/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.up_proj.weight]
Loading weights: 31%|███ | 89/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.mlp.up_proj.weight]
Loading weights: 31%|███ | 90/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.post_attention_layernorm.weight]
Loading weights: 31%|███ | 90/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.post_attention_layernorm.weight]
Loading weights: 31%|███▏ | 91/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.k_proj.bias]
Loading weights: 31%|███▏ | 91/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.k_proj.bias]
Loading weights: 32%|███▏ | 92/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.k_proj.weight]
Loading weights: 32%|███▏ | 92/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.k_proj.weight]
Loading weights: 32%|███▏ | 93/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.o_proj.weight]
Loading weights: 32%|███▏ | 93/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.o_proj.weight]
Loading weights: 32%|███▏ | 94/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.q_proj.bias]
Loading weights: 32%|███▏ | 94/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.q_proj.bias]
Loading weights: 33%|███▎ | 95/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.q_proj.weight]
Loading weights: 33%|███▎ | 95/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.q_proj.weight]
Loading weights: 33%|███▎ | 96/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.v_proj.bias]
Loading weights: 33%|███▎ | 96/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.v_proj.bias]
Loading weights: 33%|███▎ | 97/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.v_proj.weight]
Loading weights: 33%|███▎ | 97/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.7.self_attn.v_proj.weight]
Loading weights: 34%|███▍ | 98/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.input_layernorm.weight]
Loading weights: 34%|███▍ | 98/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.input_layernorm.weight]
Loading weights: 34%|███▍ | 99/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.down_proj.weight]
Loading weights: 34%|███▍ | 99/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.down_proj.weight]
Loading weights: 34%|███▍ | 100/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.gate_proj.weight]
Loading weights: 34%|███▍ | 100/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.gate_proj.weight]
Loading weights: 35%|███▍ | 101/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.up_proj.weight]
Loading weights: 35%|███▍ | 101/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.mlp.up_proj.weight]
Loading weights: 35%|███▌ | 102/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.post_attention_layernorm.weight]
Loading weights: 35%|███▌ | 102/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.post_attention_layernorm.weight]
Loading weights: 36%|███▌ | 103/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.k_proj.bias]
Loading weights: 36%|███▌ | 103/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.k_proj.bias]
Loading weights: 36%|███▌ | 104/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.k_proj.weight]
Loading weights: 36%|███▌ | 104/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.k_proj.weight]
Loading weights: 36%|███▌ | 105/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.o_proj.weight]
Loading weights: 36%|███▌ | 105/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.o_proj.weight]
Loading weights: 37%|███▋ | 106/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.q_proj.bias]
Loading weights: 37%|███▋ | 106/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.q_proj.bias]
Loading weights: 37%|███▋ | 107/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.q_proj.weight]
Loading weights: 37%|███▋ | 107/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.q_proj.weight]
Loading weights: 37%|███▋ | 108/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.v_proj.bias]
Loading weights: 37%|███▋ | 108/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.v_proj.bias]
Loading weights: 38%|███▊ | 109/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.v_proj.weight]
Loading weights: 38%|███▊ | 109/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.8.self_attn.v_proj.weight]
Loading weights: 38%|███▊ | 110/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.input_layernorm.weight]
Loading weights: 38%|███▊ | 110/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.input_layernorm.weight]
Loading weights: 38%|███▊ | 111/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.down_proj.weight]
Loading weights: 38%|███▊ | 111/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.down_proj.weight]
Loading weights: 39%|███▊ | 112/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.gate_proj.weight]
Loading weights: 39%|███▊ | 112/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.gate_proj.weight]
Loading weights: 39%|███▉ | 113/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.up_proj.weight]
Loading weights: 39%|███▉ | 113/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.mlp.up_proj.weight]
Loading weights: 39%|███▉ | 114/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.post_attention_layernorm.weight]
Loading weights: 39%|███▉ | 114/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.post_attention_layernorm.weight]
Loading weights: 40%|███▉ | 115/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.k_proj.bias]
Loading weights: 40%|███▉ | 115/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.k_proj.bias]
Loading weights: 40%|████ | 116/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.k_proj.weight]
Loading weights: 40%|████ | 116/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.k_proj.weight]
Loading weights: 40%|████ | 117/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.o_proj.weight]
Loading weights: 40%|████ | 117/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.o_proj.weight]
Loading weights: 41%|████ | 118/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.q_proj.bias]
Loading weights: 41%|████ | 118/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.q_proj.bias]
Loading weights: 41%|████ | 119/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.q_proj.weight]
Loading weights: 41%|████ | 119/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.q_proj.weight]
Loading weights: 41%|████▏ | 120/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.v_proj.bias]
Loading weights: 41%|████▏ | 120/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.v_proj.bias]
Loading weights: 42%|████▏ | 121/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.v_proj.weight]
Loading weights: 42%|████▏ | 121/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.9.self_attn.v_proj.weight]
Loading weights: 42%|████▏ | 122/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.input_layernorm.weight]
Loading weights: 42%|████▏ | 122/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.input_layernorm.weight]
Loading weights: 42%|████▏ | 123/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.down_proj.weight]
Loading weights: 42%|████▏ | 123/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.down_proj.weight]
Loading weights: 43%|████▎ | 124/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.gate_proj.weight]
Loading weights: 43%|████▎ | 124/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.gate_proj.weight]
Loading weights: 43%|████▎ | 125/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.up_proj.weight]
Loading weights: 43%|████▎ | 125/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.mlp.up_proj.weight]
Loading weights: 43%|████▎ | 126/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.post_attention_layernorm.weight]
Loading weights: 43%|████▎ | 126/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.post_attention_layernorm.weight]
Loading weights: 44%|████▍ | 127/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.k_proj.bias]
Loading weights: 44%|████▍ | 127/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.k_proj.bias]
Loading weights: 44%|████▍ | 128/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.k_proj.weight]
Loading weights: 44%|████▍ | 128/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.k_proj.weight]
Loading weights: 44%|████▍ | 129/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.o_proj.weight]
Loading weights: 44%|████▍ | 129/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.o_proj.weight]
Loading weights: 45%|████▍ | 130/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.q_proj.bias]
Loading weights: 45%|████▍ | 130/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.q_proj.bias]
Loading weights: 45%|████▌ | 131/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.q_proj.weight]
Loading weights: 45%|████▌ | 131/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.q_proj.weight]
Loading weights: 46%|████▌ | 132/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.v_proj.bias]
Loading weights: 46%|████▌ | 132/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.v_proj.bias]
Loading weights: 46%|████▌ | 133/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.v_proj.weight]
Loading weights: 46%|████▌ | 133/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.10.self_attn.v_proj.weight]
Loading weights: 46%|████▌ | 134/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.input_layernorm.weight]
Loading weights: 46%|████▌ | 134/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.input_layernorm.weight]
Loading weights: 47%|████▋ | 135/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.down_proj.weight]
Loading weights: 47%|████▋ | 135/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.down_proj.weight]
Loading weights: 47%|████▋ | 136/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.gate_proj.weight]
Loading weights: 47%|████▋ | 136/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.gate_proj.weight]
Loading weights: 47%|████▋ | 137/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.up_proj.weight]
Loading weights: 47%|████▋ | 137/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.mlp.up_proj.weight]
Loading weights: 48%|████▊ | 138/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.post_attention_layernorm.weight]
Loading weights: 48%|████▊ | 138/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.post_attention_layernorm.weight]
Loading weights: 48%|████▊ | 139/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.k_proj.bias]
Loading weights: 48%|████▊ | 139/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.k_proj.bias]
Loading weights: 48%|████▊ | 140/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.k_proj.weight]
Loading weights: 48%|████▊ | 140/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.k_proj.weight]
Loading weights: 49%|████▊ | 141/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.o_proj.weight]
Loading weights: 49%|████▊ | 141/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.o_proj.weight]
Loading weights: 49%|████▉ | 142/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.q_proj.bias]
Loading weights: 49%|████▉ | 142/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.q_proj.bias]
Loading weights: 49%|████▉ | 143/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.q_proj.weight]
Loading weights: 49%|████▉ | 143/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.q_proj.weight]
Loading weights: 50%|████▉ | 144/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.v_proj.bias]
Loading weights: 50%|████▉ | 144/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.v_proj.bias]
Loading weights: 50%|█████ | 145/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.v_proj.weight]
Loading weights: 50%|█████ | 145/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.11.self_attn.v_proj.weight]
Loading weights: 50%|█████ | 146/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.input_layernorm.weight]
Loading weights: 50%|█████ | 146/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.input_layernorm.weight]
Loading weights: 51%|█████ | 147/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.mlp.down_proj.weight]
Loading weights: 51%|█████ | 147/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.mlp.down_proj.weight]
Loading weights: 51%|█████ | 148/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.mlp.gate_proj.weight]
Loading weights: 51%|█████ | 148/290 [00:00<00:00, 555.05it/s, Materializing param=model.layers.12.mlp.gate_proj.weight]
Loading weights: 51%|█████▏ | 149/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.mlp.gate_proj.weight]
Loading weights: 51%|█████▏ | 149/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.mlp.up_proj.weight]
Loading weights: 51%|█████▏ | 149/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.mlp.up_proj.weight]
Loading weights: 52%|█████▏ | 150/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.post_attention_layernorm.weight]
Loading weights: 52%|█████▏ | 150/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.post_attention_layernorm.weight]
Loading weights: 52%|█████▏ | 151/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.k_proj.bias]
Loading weights: 52%|█████▏ | 151/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.k_proj.bias]
Loading weights: 52%|█████▏ | 152/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.k_proj.weight]
Loading weights: 52%|█████▏ | 152/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.k_proj.weight]
Loading weights: 53%|█████▎ | 153/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.o_proj.weight]
Loading weights: 53%|█████▎ | 153/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.o_proj.weight]
Loading weights: 53%|█████▎ | 154/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.q_proj.bias]
Loading weights: 53%|█████▎ | 154/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.q_proj.bias]
Loading weights: 53%|█████▎ | 155/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.q_proj.weight]
Loading weights: 53%|█████▎ | 155/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.q_proj.weight]
Loading weights: 54%|█████▍ | 156/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.v_proj.bias]
Loading weights: 54%|█████▍ | 156/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.v_proj.bias]
Loading weights: 54%|█████▍ | 157/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.v_proj.weight]
Loading weights: 54%|█████▍ | 157/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.12.self_attn.v_proj.weight]
Loading weights: 54%|█████▍ | 158/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.input_layernorm.weight]
Loading weights: 54%|█████▍ | 158/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.input_layernorm.weight]
Loading weights: 55%|█████▍ | 159/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.down_proj.weight]
Loading weights: 55%|█████▍ | 159/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.down_proj.weight]
Loading weights: 55%|█████▌ | 160/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.gate_proj.weight]
Loading weights: 55%|█████▌ | 160/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.gate_proj.weight]
Loading weights: 56%|█████▌ | 161/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.up_proj.weight]
Loading weights: 56%|█████▌ | 161/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.mlp.up_proj.weight]
Loading weights: 56%|█████▌ | 162/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.post_attention_layernorm.weight]
Loading weights: 56%|█████▌ | 162/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.post_attention_layernorm.weight]
Loading weights: 56%|█████▌ | 163/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.k_proj.bias]
Loading weights: 56%|█████▌ | 163/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.k_proj.bias]
Loading weights: 57%|█████▋ | 164/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.k_proj.weight]
Loading weights: 57%|█████▋ | 164/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.k_proj.weight]
Loading weights: 57%|█████▋ | 165/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.o_proj.weight]
Loading weights: 57%|█████▋ | 165/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.o_proj.weight]
Loading weights: 57%|█████▋ | 166/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.q_proj.bias]
Loading weights: 57%|█████▋ | 166/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.q_proj.bias]
Loading weights: 58%|█████▊ | 167/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.q_proj.weight]
Loading weights: 58%|█████▊ | 167/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.q_proj.weight]
Loading weights: 58%|█████▊ | 168/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.v_proj.bias]
Loading weights: 58%|█████▊ | 168/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.v_proj.bias]
Loading weights: 58%|█████▊ | 169/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.v_proj.weight]
Loading weights: 58%|█████▊ | 169/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.13.self_attn.v_proj.weight]
Loading weights: 59%|█████▊ | 170/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.input_layernorm.weight]
Loading weights: 59%|█████▊ | 170/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.input_layernorm.weight]
Loading weights: 59%|█████▉ | 171/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.down_proj.weight]
Loading weights: 59%|█████▉ | 171/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.down_proj.weight]
Loading weights: 59%|█████▉ | 172/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.gate_proj.weight]
Loading weights: 59%|█████▉ | 172/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.gate_proj.weight]
Loading weights: 60%|█████▉ | 173/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.up_proj.weight]
Loading weights: 60%|█████▉ | 173/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.mlp.up_proj.weight]
Loading weights: 60%|██████ | 174/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.post_attention_layernorm.weight]
Loading weights: 60%|██████ | 174/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.post_attention_layernorm.weight]
Loading weights: 60%|██████ | 175/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.k_proj.bias]
Loading weights: 60%|██████ | 175/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.k_proj.bias]
Loading weights: 61%|██████ | 176/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.k_proj.weight]
Loading weights: 61%|██████ | 176/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.k_proj.weight]
Loading weights: 61%|██████ | 177/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.o_proj.weight]
Loading weights: 61%|██████ | 177/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.o_proj.weight]
Loading weights: 61%|██████▏ | 178/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.q_proj.bias]
Loading weights: 61%|██████▏ | 178/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.q_proj.bias]
Loading weights: 62%|██████▏ | 179/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.q_proj.weight]
Loading weights: 62%|██████▏ | 179/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.q_proj.weight]
Loading weights: 62%|██████▏ | 180/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.v_proj.bias]
Loading weights: 62%|██████▏ | 180/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.v_proj.bias]
Loading weights: 62%|██████▏ | 181/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.v_proj.weight]
Loading weights: 62%|██████▏ | 181/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.14.self_attn.v_proj.weight]
Loading weights: 63%|██████▎ | 182/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.input_layernorm.weight]
Loading weights: 63%|██████▎ | 182/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.input_layernorm.weight]
Loading weights: 63%|██████▎ | 183/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.down_proj.weight]
Loading weights: 63%|██████▎ | 183/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.down_proj.weight]
Loading weights: 63%|██████▎ | 184/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.gate_proj.weight]
Loading weights: 63%|██████▎ | 184/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.gate_proj.weight]
Loading weights: 64%|██████▍ | 185/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.up_proj.weight]
Loading weights: 64%|██████▍ | 185/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.mlp.up_proj.weight]
Loading weights: 64%|██████▍ | 186/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.post_attention_layernorm.weight]
Loading weights: 64%|██████▍ | 186/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.post_attention_layernorm.weight]
Loading weights: 64%|██████▍ | 187/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.k_proj.bias]
Loading weights: 64%|██████▍ | 187/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.k_proj.bias]
Loading weights: 65%|██████▍ | 188/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.k_proj.weight]
Loading weights: 65%|██████▍ | 188/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.k_proj.weight]
Loading weights: 65%|██████▌ | 189/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.o_proj.weight]
Loading weights: 65%|██████▌ | 189/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.o_proj.weight]
Loading weights: 66%|██████▌ | 190/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.q_proj.bias]
Loading weights: 66%|██████▌ | 190/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.q_proj.bias]
Loading weights: 66%|██████▌ | 191/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.q_proj.weight]
Loading weights: 66%|██████▌ | 191/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.q_proj.weight]
Loading weights: 66%|██████▌ | 192/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.v_proj.bias]
Loading weights: 66%|██████▌ | 192/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.v_proj.bias]
Loading weights: 67%|██████▋ | 193/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.v_proj.weight]
Loading weights: 67%|██████▋ | 193/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.15.self_attn.v_proj.weight]
Loading weights: 67%|██████▋ | 194/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.input_layernorm.weight]
Loading weights: 67%|██████▋ | 194/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.input_layernorm.weight]
Loading weights: 67%|██████▋ | 195/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.down_proj.weight]
Loading weights: 67%|██████▋ | 195/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.down_proj.weight]
Loading weights: 68%|██████▊ | 196/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.gate_proj.weight]
Loading weights: 68%|██████▊ | 196/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.gate_proj.weight]
Loading weights: 68%|██████▊ | 197/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.up_proj.weight]
Loading weights: 68%|██████▊ | 197/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.mlp.up_proj.weight]
Loading weights: 68%|██████▊ | 198/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.post_attention_layernorm.weight]
Loading weights: 68%|██████▊ | 198/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.post_attention_layernorm.weight]
Loading weights: 69%|██████▊ | 199/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.k_proj.bias]
Loading weights: 69%|██████▊ | 199/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.k_proj.bias]
Loading weights: 69%|██████▉ | 200/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.k_proj.weight]
Loading weights: 69%|██████▉ | 200/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.k_proj.weight]
Loading weights: 69%|██████▉ | 201/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.o_proj.weight]
Loading weights: 69%|██████▉ | 201/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.o_proj.weight]
Loading weights: 70%|██████▉ | 202/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.q_proj.bias]
Loading weights: 70%|██████▉ | 202/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.q_proj.bias]
Loading weights: 70%|███████ | 203/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.q_proj.weight]
Loading weights: 70%|███████ | 203/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.q_proj.weight]
Loading weights: 70%|███████ | 204/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.v_proj.bias]
Loading weights: 70%|███████ | 204/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.v_proj.bias]
Loading weights: 71%|███████ | 205/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.v_proj.weight]
Loading weights: 71%|███████ | 205/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.16.self_attn.v_proj.weight]
Loading weights: 71%|███████ | 206/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.input_layernorm.weight]
Loading weights: 71%|███████ | 206/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.input_layernorm.weight]
Loading weights: 71%|███████▏ | 207/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.down_proj.weight]
Loading weights: 71%|███████▏ | 207/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.down_proj.weight]
Loading weights: 72%|███████▏ | 208/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.gate_proj.weight]
Loading weights: 72%|███████▏ | 208/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.gate_proj.weight]
Loading weights: 72%|███████▏ | 209/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.up_proj.weight]
Loading weights: 72%|███████▏ | 209/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.mlp.up_proj.weight]
Loading weights: 72%|███████▏ | 210/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.post_attention_layernorm.weight]
Loading weights: 72%|███████▏ | 210/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.post_attention_layernorm.weight]
Loading weights: 73%|███████▎ | 211/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.k_proj.bias]
Loading weights: 73%|███████▎ | 211/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.k_proj.bias]
Loading weights: 73%|███████▎ | 212/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.k_proj.weight]
Loading weights: 73%|███████▎ | 212/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.k_proj.weight]
Loading weights: 73%|███████▎ | 213/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.o_proj.weight]
Loading weights: 73%|███████▎ | 213/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.o_proj.weight]
Loading weights: 74%|███████▍ | 214/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.q_proj.bias]
Loading weights: 74%|███████▍ | 214/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.q_proj.bias]
Loading weights: 74%|███████▍ | 215/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.q_proj.weight]
Loading weights: 74%|███████▍ | 215/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.q_proj.weight]
Loading weights: 74%|███████▍ | 216/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.v_proj.bias]
Loading weights: 74%|███████▍ | 216/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.v_proj.bias]
Loading weights: 75%|███████▍ | 217/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.v_proj.weight]
Loading weights: 75%|███████▍ | 217/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.17.self_attn.v_proj.weight]
Loading weights: 75%|███████▌ | 218/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.input_layernorm.weight]
Loading weights: 75%|███████▌ | 218/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.input_layernorm.weight]
Loading weights: 76%|███████▌ | 219/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.down_proj.weight]
Loading weights: 76%|███████▌ | 219/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.down_proj.weight]
Loading weights: 76%|███████▌ | 220/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.gate_proj.weight]
Loading weights: 76%|███████▌ | 220/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.gate_proj.weight]
Loading weights: 76%|███████▌ | 221/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.up_proj.weight]
Loading weights: 76%|███████▌ | 221/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.mlp.up_proj.weight]
Loading weights: 77%|███████▋ | 222/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.post_attention_layernorm.weight]
Loading weights: 77%|███████▋ | 222/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.post_attention_layernorm.weight]
Loading weights: 77%|███████▋ | 223/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.k_proj.bias]
Loading weights: 77%|███████▋ | 223/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.k_proj.bias]
Loading weights: 77%|███████▋ | 224/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.k_proj.weight]
Loading weights: 77%|███████▋ | 224/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.k_proj.weight]
Loading weights: 78%|███████▊ | 225/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.o_proj.weight]
Loading weights: 78%|███████▊ | 225/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.o_proj.weight]
Loading weights: 78%|███████▊ | 226/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.q_proj.bias]
Loading weights: 78%|███████▊ | 226/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.q_proj.bias]
Loading weights: 78%|███████▊ | 227/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.q_proj.weight]
Loading weights: 78%|███████▊ | 227/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.q_proj.weight]
Loading weights: 79%|███████▊ | 228/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.v_proj.bias]
Loading weights: 79%|███████▊ | 228/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.v_proj.bias]
Loading weights: 79%|███████▉ | 229/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.v_proj.weight]
Loading weights: 79%|███████▉ | 229/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.18.self_attn.v_proj.weight]
Loading weights: 79%|███████▉ | 230/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.input_layernorm.weight]
Loading weights: 79%|███████▉ | 230/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.input_layernorm.weight]
Loading weights: 80%|███████▉ | 231/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.down_proj.weight]
Loading weights: 80%|███████▉ | 231/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.down_proj.weight]
Loading weights: 80%|████████ | 232/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.gate_proj.weight]
Loading weights: 80%|████████ | 232/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.gate_proj.weight]
Loading weights: 80%|████████ | 233/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.up_proj.weight]
Loading weights: 80%|████████ | 233/290 [00:00<00:00, 763.57it/s, Materializing param=model.layers.19.mlp.up_proj.weight]
Loading weights: 81%|████████ | 234/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.mlp.up_proj.weight]
Loading weights: 81%|████████ | 234/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.post_attention_layernorm.weight]
Loading weights: 81%|████████ | 234/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.post_attention_layernorm.weight]
Loading weights: 81%|████████ | 235/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.k_proj.bias]
Loading weights: 81%|████████ | 235/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.k_proj.bias]
Loading weights: 81%|████████▏ | 236/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.k_proj.weight]
Loading weights: 81%|████████▏ | 236/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.k_proj.weight]
Loading weights: 82%|████████▏ | 237/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.o_proj.weight]
Loading weights: 82%|████████▏ | 237/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.o_proj.weight]
Loading weights: 82%|████████▏ | 238/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.q_proj.bias]
Loading weights: 82%|████████▏ | 238/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.q_proj.bias]
Loading weights: 82%|████████▏ | 239/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.q_proj.weight]
Loading weights: 82%|████████▏ | 239/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.q_proj.weight]
Loading weights: 83%|████████▎ | 240/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.v_proj.bias]
Loading weights: 83%|████████▎ | 240/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.v_proj.bias]
Loading weights: 83%|████████▎ | 241/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.v_proj.weight]
Loading weights: 83%|████████▎ | 241/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.19.self_attn.v_proj.weight]
Loading weights: 83%|████████▎ | 242/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.input_layernorm.weight]
Loading weights: 83%|████████▎ | 242/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.input_layernorm.weight]
Loading weights: 84%|████████▍ | 243/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.down_proj.weight]
Loading weights: 84%|████████▍ | 243/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.down_proj.weight]
Loading weights: 84%|████████▍ | 244/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.gate_proj.weight]
Loading weights: 84%|████████▍ | 244/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.gate_proj.weight]
Loading weights: 84%|████████▍ | 245/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.up_proj.weight]
Loading weights: 84%|████████▍ | 245/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.mlp.up_proj.weight]
Loading weights: 85%|████████▍ | 246/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.post_attention_layernorm.weight]
Loading weights: 85%|████████▍ | 246/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.post_attention_layernorm.weight]
Loading weights: 85%|████████▌ | 247/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.k_proj.bias]
Loading weights: 85%|████████▌ | 247/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.k_proj.bias]
Loading weights: 86%|████████▌ | 248/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.k_proj.weight]
Loading weights: 86%|████████▌ | 248/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.k_proj.weight]
Loading weights: 86%|████████▌ | 249/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.o_proj.weight]
Loading weights: 86%|████████▌ | 249/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.o_proj.weight]
Loading weights: 86%|████████▌ | 250/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.q_proj.bias]
Loading weights: 86%|████████▌ | 250/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.q_proj.bias]
Loading weights: 87%|████████▋ | 251/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.q_proj.weight]
Loading weights: 87%|████████▋ | 251/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.q_proj.weight]
Loading weights: 87%|████████▋ | 252/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.v_proj.bias]
Loading weights: 87%|████████▋ | 252/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.v_proj.bias]
Loading weights: 87%|████████▋ | 253/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.v_proj.weight]
Loading weights: 87%|████████▋ | 253/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.20.self_attn.v_proj.weight]
Loading weights: 88%|████████▊ | 254/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.input_layernorm.weight]
Loading weights: 88%|████████▊ | 254/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.input_layernorm.weight]
Loading weights: 88%|████████▊ | 255/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.down_proj.weight]
Loading weights: 88%|████████▊ | 255/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.down_proj.weight]
Loading weights: 88%|████████▊ | 256/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.gate_proj.weight]
Loading weights: 88%|████████▊ | 256/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.gate_proj.weight]
Loading weights: 89%|████████▊ | 257/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.up_proj.weight]
Loading weights: 89%|████████▊ | 257/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.mlp.up_proj.weight]
Loading weights: 89%|████████▉ | 258/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.post_attention_layernorm.weight]
Loading weights: 89%|████████▉ | 258/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.post_attention_layernorm.weight]
Loading weights: 89%|████████▉ | 259/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.k_proj.bias]
Loading weights: 89%|████████▉ | 259/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.k_proj.bias]
Loading weights: 90%|████████▉ | 260/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.k_proj.weight]
Loading weights: 90%|████████▉ | 260/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.k_proj.weight]
Loading weights: 90%|█████████ | 261/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.o_proj.weight]
Loading weights: 90%|█████████ | 261/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.o_proj.weight]
Loading weights: 90%|█████████ | 262/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.q_proj.bias]
Loading weights: 90%|█████████ | 262/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.q_proj.bias]
Loading weights: 91%|█████████ | 263/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.q_proj.weight]
Loading weights: 91%|█████████ | 263/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.q_proj.weight]
Loading weights: 91%|█████████ | 264/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.v_proj.bias]
Loading weights: 91%|█████████ | 264/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.v_proj.bias]
Loading weights: 91%|█████████▏| 265/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.v_proj.weight]
Loading weights: 91%|█████████▏| 265/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.21.self_attn.v_proj.weight]
Loading weights: 92%|█████████▏| 266/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.input_layernorm.weight]
Loading weights: 92%|█████████▏| 266/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.input_layernorm.weight]
Loading weights: 92%|█████████▏| 267/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.down_proj.weight]
Loading weights: 92%|█████████▏| 267/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.down_proj.weight]
Loading weights: 92%|█████████▏| 268/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.gate_proj.weight]
Loading weights: 92%|█████████▏| 268/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.gate_proj.weight]
Loading weights: 93%|█████████▎| 269/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.up_proj.weight]
Loading weights: 93%|█████████▎| 269/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.mlp.up_proj.weight]
Loading weights: 93%|█████████▎| 270/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.post_attention_layernorm.weight]
Loading weights: 93%|█████████▎| 270/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.post_attention_layernorm.weight]
Loading weights: 93%|█████████▎| 271/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.k_proj.bias]
Loading weights: 93%|█████████▎| 271/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.k_proj.bias]
Loading weights: 94%|█████████▍| 272/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.k_proj.weight]
Loading weights: 94%|█████████▍| 272/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.k_proj.weight]
Loading weights: 94%|█████████▍| 273/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.o_proj.weight]
Loading weights: 94%|█████████▍| 273/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.o_proj.weight]
Loading weights: 94%|█████████▍| 274/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.q_proj.bias]
Loading weights: 94%|█████████▍| 274/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.q_proj.bias]
Loading weights: 95%|█████████▍| 275/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.q_proj.weight]
Loading weights: 95%|█████████▍| 275/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.q_proj.weight]
Loading weights: 95%|█████████▌| 276/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.v_proj.bias]
Loading weights: 95%|█████████▌| 276/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.v_proj.bias]
Loading weights: 96%|█████████▌| 277/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.v_proj.weight]
Loading weights: 96%|█████████▌| 277/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.22.self_attn.v_proj.weight]
Loading weights: 96%|█████████▌| 278/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.input_layernorm.weight]
Loading weights: 96%|█████████▌| 278/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.input_layernorm.weight]
Loading weights: 96%|█████████▌| 279/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.down_proj.weight]
Loading weights: 96%|█████████▌| 279/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.down_proj.weight]
Loading weights: 97%|█████████▋| 280/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.gate_proj.weight]
Loading weights: 97%|█████████▋| 280/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.gate_proj.weight]
Loading weights: 97%|█████████▋| 281/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.up_proj.weight]
Loading weights: 97%|█████████▋| 281/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.mlp.up_proj.weight]
Loading weights: 97%|█████████▋| 282/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.post_attention_layernorm.weight]
Loading weights: 97%|█████████▋| 282/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.post_attention_layernorm.weight]
Loading weights: 98%|█████████▊| 283/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.k_proj.bias]
Loading weights: 98%|█████████▊| 283/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.k_proj.bias]
Loading weights: 98%|█████████▊| 284/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.k_proj.weight]
Loading weights: 98%|█████████▊| 284/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.k_proj.weight]
Loading weights: 98%|█████████▊| 285/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.o_proj.weight]
Loading weights: 98%|█████████▊| 285/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.o_proj.weight]
Loading weights: 99%|█████████▊| 286/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.q_proj.bias]
Loading weights: 99%|█████████▊| 286/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.q_proj.bias]
Loading weights: 99%|█████████▉| 287/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.q_proj.weight]
Loading weights: 99%|█████████▉| 287/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.q_proj.weight]
Loading weights: 99%|█████████▉| 288/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.v_proj.bias]
Loading weights: 99%|█████████▉| 288/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.v_proj.bias]
Loading weights: 100%|█████████▉| 289/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.v_proj.weight]
Loading weights: 100%|█████████▉| 289/290 [00:00<00:00, 791.52it/s, Materializing param=model.layers.23.self_attn.v_proj.weight]
Loading weights: 100%|██████████| 290/290 [00:00<00:00, 791.52it/s, Materializing param=model.norm.weight]
Loading weights: 100%|██████████| 290/290 [00:00<00:00, 791.52it/s, Materializing param=model.norm.weight]
Loading weights: 100%|██████████| 290/290 [00:00<00:00, 784.15it/s, Materializing param=model.norm.weight]
The tokenizer has new PAD/BOS/EOS tokens that differ from the model config and generation config. The model config and generation config were aligned accordingly, being updated with the tokenizer's values. Updated tokens: {'bos_token_id': None, 'pad_token_id': 151643}.
Model loaded: 494.0M parameters
[4] Building training dataset...
[5] Configuring GRPO trainer...
[6] Training for 250 steps...
Reward will be logged every 10 steps.
Watch for increasing reward_mean over time.
0%| | 0/250 [00:00<?, ?it/s]
0%| | 1/250 [00:19<1:20:59, 19.52s/it]
1%| | 2/250 [00:52<1:54:16, 27.65s/it]
1%| | 3/250 [01:27<2:07:17, 30.92s/it]
2%|▏ | 4/250 [02:02<2:13:52, 32.65s/it]
2%|▏ | 5/250 [02:36<2:15:20, 33.14s/it]
2%|▏ | 6/250 [03:12<2:17:48, 33.89s/it]
3%|▎ | 7/250 [03:47<2:19:03, 34.34s/it]
3%|▎ | 8/250 [04:22<2:19:12, 34.52s/it]
4%|▎ | 9/250 [04:57<2:19:24, 34.71s/it]
4%|▍ | 10/250 [05:32<2:18:48, 34.70s/it]
4%|▍ | 10/250 [05:32<2:18:48, 34.70s/it]
4%|▍ | 11/250 [06:07<2:19:10, 34.94s/it]
5%|▍ | 12/250 [06:43<2:19:03, 35.06s/it]
5%|▌ | 13/250 [07:18<2:18:59, 35.19s/it]
6%|▌ | 14/250 [07:53<2:18:29, 35.21s/it]
6%|▌ | 15/250 [08:28<2:17:46, 35.18s/it]
6%|▋ | 16/250 [09:03<2:16:30, 35.00s/it]
7%|▋ | 17/250 [09:38<2:15:43, 34.95s/it]
7%|▋ | 18/250 [10:13<2:15:07, 34.95s/it]
8%|▊ | 19/250 [10:48<2:15:00, 35.07s/it]
8%|▊ | 20/250 [11:23<2:14:24, 35.06s/it]
8%|▊ | 20/250 [11:23<2:14:24, 35.06s/it]
8%|▊ | 21/250 [11:58<2:13:46, 35.05s/it]
9%|▉ | 22/250 [12:34<2:13:27, 35.12s/it]
9%|▉ | 23/250 [13:08<2:12:30, 35.02s/it]
10%|▉ | 24/250 [13:43<2:11:38, 34.95s/it]
10%|█ | 25/250 [14:19<2:11:54, 35.18s/it]
10%|█ | 26/250 [14:54<2:11:14, 35.15s/it]
11%|█ | 27/250 [15:29<2:10:22, 35.08s/it]
11%|█ | 28/250 [16:04<2:09:38, 35.04s/it]
12%|█▏ | 29/250 [16:39<2:09:23, 35.13s/it]
12%|█▏ | 30/250 [17:14<2:08:41, 35.10s/it]
12%|█▏ | 30/250 [17:14<2:08:41, 35.10s/it]
12%|█▏ | 31/250 [17:49<2:08:03, 35.08s/it]
13%|█▎ | 32/250 [18:24<2:07:05, 34.98s/it]
13%|█▎ | 33/250 [18:59<2:06:15, 34.91s/it]
14%|█▎ | 34/250 [19:34<2:05:52, 34.96s/it]
14%|█▍ | 35/250 [20:09<2:05:14, 34.95s/it]
14%|█▍ | 36/250 [20:44<2:05:11, 35.10s/it]
15%|█▍ | 37/250 [21:20<2:04:57, 35.20s/it]
15%|█▌ | 38/250 [21:55<2:04:30, 35.24s/it]
16%|█▌ | 39/250 [22:30<2:03:19, 35.07s/it]
16%|█▌ | 40/250 [23:04<2:02:28, 34.99s/it]
16%|█▌ | 40/250 [23:04<2:02:28, 34.99s/it]
16%|█▋ | 41/250 [23:39<2:01:43, 34.95s/it]
17%|█▋ | 42/250 [24:15<2:01:36, 35.08s/it]
17%|█▋ | 43/250 [24:50<2:01:32, 35.23s/it]
18%|█▊ | 44/250 [25:25<2:00:41, 35.15s/it]
18%|█▊ | 45/250 [26:00<2:00:14, 35.19s/it]
18%|█▊ | 46/250 [26:36<1:59:31, 35.16s/it]
19%|█▉ | 47/250 [27:10<1:58:37, 35.06s/it]
19%|█▉ | 48/250 [27:46<1:58:30, 35.20s/it]
20%|█▉ | 49/250 [28:21<1:57:43, 35.14s/it]
20%|██ | 50/250 [28:56<1:57:05, 35.13s/it]
20%|██ | 50/250 [28:56<1:57:05, 35.13s/it]
20%|██ | 51/250 [29:32<1:56:55, 35.25s/it]
21%|██ | 52/250 [30:06<1:56:00, 35.15s/it]
21%|██ | 53/250 [30:41<1:54:57, 35.01s/it]
22%|██▏ | 54/250 [31:16<1:54:28, 35.04s/it]
22%|██▏ | 55/250 [31:51<1:54:00, 35.08s/it]
22%|██▏ | 56/250 [32:26<1:53:17, 35.04s/it]
23%|██▎ | 57/250 [33:02<1:53:10, 35.18s/it]
23%|██▎ | 58/250 [33:37<1:52:11, 35.06s/it]
24%|██▎ | 59/250 [34:12<1:51:27, 35.01s/it]
24%|██▍ | 60/250 [34:47<1:51:03, 35.07s/it]
24%|██▍ | 60/250 [34:47<1:51:03, 35.07s/it]
24%|██▍ | 61/250 [35:22<1:50:45, 35.16s/it]
25%|██▍ | 62/250 [35:57<1:50:09, 35.15s/it]
25%|██▌ | 63/250 [36:32<1:49:25, 35.11s/it]
26%|██▌ | 64/250 [37:07<1:48:32, 35.01s/it]
26%|██▌ | 65/250 [37:42<1:48:14, 35.11s/it]
26%|██▋ | 66/250 [38:18<1:47:48, 35.15s/it]
27%|██▋ | 67/250 [38:53<1:47:12, 35.15s/it]
27%|██▋ | 68/250 [39:27<1:46:03, 34.96s/it]
28%|██▊ | 69/250 [40:02<1:45:22, 34.93s/it]
28%|██▊ | 70/250 [40:38<1:45:10, 35.06s/it]
28%|██▊ | 70/250 [40:38<1:45:10, 35.06s/it]
28%|██▊ | 71/250 [41:12<1:44:29, 35.03s/it]
29%|██▉ | 72/250 [41:48<1:44:08, 35.10s/it]
29%|██▉ | 73/250 [42:23<1:43:40, 35.14s/it]
30%|██▉ | 74/250 [42:58<1:43:13, 35.19s/it]
30%|███ | 75/250 [43:33<1:42:18, 35.08s/it]
30%|███ | 76/250 [44:08<1:41:15, 34.92s/it]
31%|███ | 77/250 [44:43<1:40:57, 35.02s/it]
31%|███ | 78/250 [45:18<1:40:36, 35.10s/it]
32%|███▏ | 79/250 [45:53<1:39:48, 35.02s/it]
32%|███▏ | 80/250 [46:28<1:38:58, 34.93s/it]
32%|███▏ | 80/250 [46:28<1:38:58, 34.93s/it]
32%|███▏ | 81/250 [47:02<1:38:13, 34.87s/it]
33%|███▎ | 82/250 [47:37<1:37:44, 34.91s/it]
33%|███▎ | 83/250 [48:12<1:37:10, 34.91s/it]
34%|███▎ | 84/250 [48:47<1:36:32, 34.90s/it]
34%|███▍ | 85/250 [49:23<1:36:24, 35.06s/it]
34%|███▍ | 86/250 [49:58<1:35:39, 35.00s/it]
35%|███▍ | 87/250 [50:32<1:34:40, 34.85s/it]
35%|███▌ | 88/250 [51:07<1:34:15, 34.91s/it]
36%|███▌ | 89/250 [51:43<1:34:17, 35.14s/it]
36%|███▌ | 90/250 [52:18<1:33:34, 35.09s/it]
36%|███▌ | 90/250 [52:18<1:33:34, 35.09s/it]
36%|███▋ | 91/250 [52:53<1:33:16, 35.20s/it]
37%|███▋ | 92/250 [53:28<1:32:16, 35.04s/it]
37%|███▋ | 93/250 [54:03<1:31:41, 35.04s/it]
38%|███▊ | 94/250 [54:38<1:30:54, 34.97s/it]
38%|███▊ | 95/250 [55:13<1:30:44, 35.13s/it]
38%|███▊ | 96/250 [55:48<1:29:59, 35.06s/it]
39%|███▉ | 97/250 [56:23<1:29:07, 34.95s/it]
39%|███▉ | 98/250 [56:58<1:28:48, 35.05s/it]
40%|███▉ | 99/250 [57:33<1:28:00, 34.97s/it]
40%|████ | 100/250 [58:08<1:27:33, 35.02s/it]
40%|████ | 100/250 [58:08<1:27:33, 35.02s/it]{'loss': '-0.01747', 'grad_norm': '0', 'learning_rate': '1.928e-05', 'num_tokens': '6.185e+04', 'completions/mean_length': '362.8', 'completions/min_length': '318.1', 'completions/max_length': '379.3', 'completions/clipped_ratio': '0.8625', 'completions/mean_terminated_length': '46.41', 'completions/min_terminated_length': '38.1', 'completions/max_terminated_length': '57.8', 'rewards/reward_fn/mean': '-0.09879', 'rewards/reward_fn/std': '0.9303', 'reward': '-0.09879', 'reward_std': '0.9303', 'frac_reward_zero_std': '0.8', 'entropy': '0.5937', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '31.72', 'epoch': '0.3333'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.848e-05', 'num_tokens': '1.268e+05', 'completions/mean_length': '400', 'completions/min_length': '400', 'completions/max_length': '400', 'completions/clipped_ratio': '1', 'completions/mean_terminated_length': '0', 'completions/min_terminated_length': '0', 'completions/max_terminated_length': '0', 'rewards/reward_fn/mean': '0.04825', 'rewards/reward_fn/std': '0.8191', 'reward': '0.04825', 'reward_std': '0.8191', 'frac_reward_zero_std': '1', 'entropy': '0.6594', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.55', 'epoch': '0.6667'}
{'loss': '-0.005067', 'grad_norm': '1.805', 'learning_rate': '1.768e-05', 'num_tokens': '1.915e+05', 'completions/mean_length': '396.2', 'completions/min_length': '369.7', 'completions/max_length': '400', 'completions/clipped_ratio': '0.975', 'completions/mean_terminated_length': '49.7', 'completions/min_terminated_length': '49.7', 'completions/max_terminated_length': '49.7', 'rewards/reward_fn/mean': '0.1105', 'rewards/reward_fn/std': '0.7119', 'reward': '0.1105', 'reward_std': '0.7119', 'frac_reward_zero_std': '0.9', 'entropy': '0.5579', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.49', 'epoch': '1'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.688e-05', 'num_tokens': '2.565e+05', 'completions/mean_length': '400', 'completions/min_length': '400', 'completions/max_length': '400', 'completions/clipped_ratio': '1', 'completions/mean_terminated_length': '0', 'completions/min_terminated_length': '0', 'completions/max_terminated_length': '0', 'rewards/reward_fn/mean': '0.07887', 'rewards/reward_fn/std': '0.8938', 'reward': '0.07887', 'reward_std': '0.8938', 'frac_reward_zero_std': '0.95', 'entropy': '0.00434', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.44', 'epoch': '1.333'}
{'loss': '0', 'grad_norm': '0.3848', 'learning_rate': '1.608e-05', 'num_tokens': '3.211e+05', 'completions/mean_length': '395.1', 'completions/min_length': '360.9', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9875', 'completions/mean_terminated_length': '0.9', 'completions/min_terminated_length': '0.9', 'completions/max_terminated_length': '0.9', 'rewards/reward_fn/mean': '-0.1644', 'rewards/reward_fn/std': '1.064', 'reward': '-0.1644', 'reward_std': '1.064', 'frac_reward_zero_std': '0.975', 'entropy': '0.004431', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.56', 'epoch': '1.667'}
{'loss': '0.009756', 'grad_norm': '0', 'learning_rate': '1.528e-05', 'num_tokens': '3.848e+05', 'completions/mean_length': '385.5', 'completions/min_length': '283.7', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9625', 'completions/mean_terminated_length': '3.7', 'completions/min_terminated_length': '3.7', 'completions/max_terminated_length': '3.7', 'rewards/reward_fn/mean': '-0.1656', 'rewards/reward_fn/std': '0.936', 'reward': '-0.1656', 'reward_std': '0.936', 'frac_reward_zero_std': '0.925', 'entropy': '0.09349', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.49', 'epoch': '2'}
{'loss': '-0.01167', 'grad_norm': '2.406', 'learning_rate': '1.448e-05', 'num_tokens': '4.458e+05', 'completions/mean_length': '351.8', 'completions/min_length': '159.8', 'completions/max_length': '400', 'completions/clipped_ratio': '0.8625', 'completions/mean_terminated_length': '42.27', 'completions/min_terminated_length': '39.8', 'completions/max_terminated_length': '44.3', 'rewards/reward_fn/mean': '0.08913', 'rewards/reward_fn/std': '0.9902', 'reward': '0.08913', 'reward_std': '0.9902', 'frac_reward_zero_std': '0.925', 'entropy': '0.6014', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.48', 'epoch': '2.333'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.368e-05', 'num_tokens': '5.079e+05', 'completions/mean_length': '365.5', 'completions/min_length': '192.7', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9', 'completions/mean_terminated_length': '34.75', 'completions/min_terminated_length': '32.7', 'completions/max_terminated_length': '36.8', 'rewards/reward_fn/mean': '-0.1438', 'rewards/reward_fn/std': '0.6797', 'reward': '-0.1438', 'reward_std': '0.6797', 'frac_reward_zero_std': '0.975', 'entropy': '0.7648', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.44', 'epoch': '2.667'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.288e-05', 'num_tokens': '5.719e+05', 'completions/mean_length': '386.4', 'completions/min_length': '312.1', 'completions/max_length': '400', 'completions/clipped_ratio': '0.95', 'completions/mean_terminated_length': '35.4', 'completions/min_terminated_length': '32.1', 'completions/max_terminated_length': '38.7', 'rewards/reward_fn/mean': '-0.091', 'rewards/reward_fn/std': '0.7491', 'reward': '-0.091', 'reward_std': '0.7491', 'frac_reward_zero_std': '1', 'entropy': '0.8177', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.42', 'epoch': '3'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.208e-05', 'num_tokens': '6.358e+05', 'completions/mean_length': '386.5', 'completions/min_length': '292.1', 'completions/max_length': '400', 'completions/clipped_ratio': '0.95', 'completions/mean_terminated_length': '52.1', 'completions/min_terminated_length': '52.1', 'completions/max_terminated_length': '52.1', 'rewards/reward_fn/mean': '-0.1185', 'rewards/reward_fn/std': '0.978', 'reward': '-0.1185', 'reward_std': '0.978', 'frac_reward_zero_std': '1', 'entropy': '0.7871', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.45', 'epoch': '3.333'}
Writing model shards: 0%| | 0/1 [00:00<?, ?it/s]
Writing model shards: 100%|██████████| 1/1 [00:08<00:00, 8.32s/it]
Writing model shards: 100%|██████████| 1/1 [00:08<00:00, 8.32s/it]
40%|████ | 101/250 [59:05<1:43:00, 41.48s/it]
41%|████ | 102/250 [59:40<1:37:45, 39.63s/it]
41%|████ | 103/250 [1:00:15<1:33:26, 38.14s/it]
42%|████▏ | 104/250 [1:00:50<1:30:29, 37.19s/it]
42%|████▏ | 105/250 [1:01:24<1:28:07, 36.46s/it]
42%|████▏ | 106/250 [1:01:59<1:26:11, 35.91s/it]
43%|████▎ | 107/250 [1:02:34<1:24:45, 35.56s/it]
43%|████▎ | 108/250 [1:03:09<1:23:45, 35.39s/it]
44%|████▎ | 109/250 [1:03:44<1:23:08, 35.38s/it]
44%|████▍ | 110/250 [1:04:19<1:22:30, 35.36s/it]
44%|████▍ | 110/250 [1:04:19<1:22:30, 35.36s/it]
44%|████▍ | 111/250 [1:04:55<1:21:56, 35.37s/it]
45%|████▍ | 112/250 [1:05:29<1:20:50, 35.15s/it]
45%|████▌ | 113/250 [1:06:04<1:20:09, 35.10s/it]
46%|████▌ | 114/250 [1:06:39<1:19:30, 35.07s/it]
46%|████▌ | 115/250 [1:07:15<1:19:08, 35.18s/it]
46%|████▋ | 116/250 [1:07:50<1:18:22, 35.09s/it]
47%|████▋ | 117/250 [1:08:25<1:17:54, 35.14s/it]
47%|████▋ | 118/250 [1:09:00<1:17:10, 35.08s/it]
48%|████▊ | 119/250 [1:09:35<1:16:49, 35.19s/it]
48%|████▊ | 120/250 [1:10:10<1:15:50, 35.00s/it]
48%|████▊ | 120/250 [1:10:10<1:15:50, 35.00s/it]
48%|████▊ | 121/250 [1:10:45<1:15:15, 35.00s/it]
49%|████▉ | 122/250 [1:11:20<1:14:59, 35.16s/it]
49%|████▉ | 123/250 [1:11:55<1:14:21, 35.13s/it]
50%|████▉ | 124/250 [1:12:31<1:13:45, 35.12s/it]
50%|█████ | 125/250 [1:13:05<1:12:59, 35.04s/it]
50%|█████ | 126/250 [1:13:41<1:12:30, 35.09s/it]
51%|█████ | 127/250 [1:14:16<1:11:54, 35.08s/it]
51%|█████ | 128/250 [1:14:51<1:11:28, 35.15s/it]
52%|█████▏ | 129/250 [1:15:26<1:10:52, 35.14s/it]
52%|█████▏ | 130/250 [1:16:01<1:10:00, 35.00s/it]
52%|█████▏ | 130/250 [1:16:01<1:10:00, 35.00s/it]
52%|█████▏ | 131/250 [1:16:36<1:09:35, 35.09s/it]
53%|█████▎ | 132/250 [1:17:11<1:08:41, 34.93s/it]
53%|█████▎ | 133/250 [1:17:46<1:08:14, 35.00s/it]
54%|█████▎ | 134/250 [1:18:21<1:07:43, 35.03s/it]
54%|█████▍ | 135/250 [1:18:56<1:07:24, 35.17s/it]
54%|█████▍ | 136/250 [1:19:31<1:06:42, 35.11s/it]
55%|█████▍ | 137/250 [1:20:06<1:05:54, 34.99s/it]
55%|█████▌ | 138/250 [1:20:41<1:05:28, 35.08s/it]
56%|█████▌ | 139/250 [1:21:17<1:05:06, 35.20s/it]
56%|█████▌ | 140/250 [1:21:52<1:04:23, 35.12s/it]
56%|█████▌ | 140/250 [1:21:52<1:04:23, 35.12s/it]
56%|█████▋ | 141/250 [1:22:27<1:03:54, 35.18s/it]
57%|█████▋ | 142/250 [1:23:02<1:03:25, 35.24s/it]
57%|█████▋ | 143/250 [1:23:37<1:02:38, 35.13s/it]
58%|█████▊ | 144/250 [1:24:12<1:01:53, 35.03s/it]
58%|█████▊ | 145/250 [1:24:47<1:01:18, 35.03s/it]
58%|█████▊ | 146/250 [1:25:22<1:00:39, 34.99s/it]
59%|█████▉ | 147/250 [1:25:58<1:00:19, 35.14s/it]
59%|█████▉ | 148/250 [1:26:32<59:32, 35.03s/it]
60%|█████▉ | 149/250 [1:27:07<58:56, 35.02s/it]
60%|██████ | 150/250 [1:27:42<58:22, 35.02s/it]
60%|██████ | 150/250 [1:27:42<58:22, 35.02s/it]
60%|██████ | 151/250 [1:28:18<57:51, 35.06s/it]
61%|██████ | 152/250 [1:28:52<57:11, 35.01s/it]
61%|██████ | 153/250 [1:29:27<56:27, 34.92s/it]
62%|██████▏ | 154/250 [1:30:02<55:53, 34.93s/it]
62%|██████▏ | 155/250 [1:30:37<55:10, 34.85s/it]
62%|██████▏ | 156/250 [1:31:12<54:37, 34.87s/it]
63%|██████▎ | 157/250 [1:31:47<54:25, 35.12s/it]
63%|██████▎ | 158/250 [1:32:22<53:42, 35.03s/it]
64%|██████▎ | 159/250 [1:32:57<53:14, 35.11s/it]
64%|██████▍ | 160/250 [1:33:32<52:31, 35.02s/it]
64%|██████▍ | 160/250 [1:33:32<52:31, 35.02s/it]
64%|██████▍ | 161/250 [1:34:08<52:06, 35.13s/it]
65%|██████▍ | 162/250 [1:34:42<51:22, 35.03s/it]
65%|██████▌ | 163/250 [1:35:17<50:39, 34.93s/it]
66%|██████▌ | 164/250 [1:35:52<50:13, 35.04s/it]
66%|██████▌ | 165/250 [1:36:27<49:31, 34.96s/it]
66%|██████▋ | 166/250 [1:37:03<49:05, 35.06s/it]
67%|██████▋ | 167/250 [1:37:37<48:27, 35.03s/it]
67%|██████▋ | 168/250 [1:38:13<47:56, 35.08s/it]
68%|██████▊ | 169/250 [1:38:48<47:24, 35.12s/it]
68%|██████▊ | 170/250 [1:39:23<46:48, 35.10s/it]
68%|██████▊ | 170/250 [1:39:23<46:48, 35.10s/it]
68%|██████▊ | 171/250 [1:39:58<46:19, 35.18s/it]
69%|██████▉ | 172/250 [1:40:33<45:34, 35.05s/it]
69%|██████▉ | 173/250 [1:41:09<45:08, 35.18s/it]
70%|██████▉ | 174/250 [1:41:43<44:25, 35.07s/it]
70%|███████ | 175/250 [1:42:19<43:52, 35.10s/it]
70%|███████ | 176/250 [1:42:54<43:16, 35.09s/it]
71%|███████ | 177/250 [1:43:29<42:39, 35.06s/it]
71%|███████ | 178/250 [1:44:04<42:03, 35.06s/it]
72%|███████▏ | 179/250 [1:44:39<41:33, 35.13s/it]
72%|███████▏ | 180/250 [1:45:14<40:48, 34.98s/it]
72%|███████▏ | 180/250 [1:45:14<40:48, 34.98s/it]
72%|███████▏ | 181/250 [1:45:49<40:15, 35.01s/it]
73%|███████▎ | 182/250 [1:46:23<39:37, 34.97s/it]
73%|███████▎ | 183/250 [1:46:58<38:59, 34.92s/it]
74%|███████▎ | 184/250 [1:47:33<38:27, 34.97s/it]
74%|███████▍ | 185/250 [1:48:09<38:01, 35.10s/it]
74%|███████▍ | 186/250 [1:48:44<37:22, 35.03s/it]
75%|███████▍ | 187/250 [1:49:19<36:46, 35.03s/it]
75%|███████▌ | 188/250 [1:49:54<36:08, 34.98s/it]
76%|███████▌ | 189/250 [1:50:29<35:37, 35.04s/it]
76%|███████▌ | 190/250 [1:51:04<34:58, 34.98s/it]
76%|███████▌ | 190/250 [1:51:04<34:58, 34.98s/it]
76%|███████▋ | 191/250 [1:51:38<34:20, 34.93s/it]
77%|███████▋ | 192/250 [1:52:13<33:40, 34.83s/it]
77%|███████▋ | 193/250 [1:52:49<33:20, 35.10s/it]
78%|███████▊ | 194/250 [1:53:24<32:43, 35.06s/it]
78%|███████▊ | 195/250 [1:53:59<32:06, 35.04s/it]
78%|███████▊ | 196/250 [1:54:33<31:26, 34.94s/it]
79%|███████▉ | 197/250 [1:55:09<30:58, 35.06s/it]
79%|███████▉ | 198/250 [1:55:43<30:15, 34.91s/it]
80%|███████▉ | 199/250 [1:56:18<29:42, 34.95s/it]
80%|████████ | 200/250 [1:56:54<29:12, 35.04s/it]
80%|████████ | 200/250 [1:56:54<29:12, 35.04s/it]{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.128e-05', 'num_tokens': '6.999e+05', 'completions/mean_length': '389.8', 'completions/min_length': '336.6', 'completions/max_length': '400', 'completions/clipped_ratio': '0.95', 'completions/mean_terminated_length': '65.95', 'completions/min_terminated_length': '56.6', 'completions/max_terminated_length': '75.3', 'rewards/reward_fn/mean': '0.06825', 'rewards/reward_fn/std': '0.8598', 'reward': '0.06825', 'reward_std': '0.8598', 'frac_reward_zero_std': '1', 'entropy': '0.8245', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.43', 'epoch': '3.667'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.048e-05', 'num_tokens': '7.647e+05', 'completions/mean_length': '398.6', 'completions/min_length': '389', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9875', 'completions/mean_terminated_length': '29', 'completions/min_terminated_length': '29', 'completions/max_terminated_length': '29', 'rewards/reward_fn/mean': '-0.05875', 'rewards/reward_fn/std': '0.7074', 'reward': '-0.05875', 'reward_std': '0.7074', 'frac_reward_zero_std': '1', 'entropy': '0.8096', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.48', 'epoch': '4'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '9.68e-06', 'num_tokens': '8.289e+05', 'completions/mean_length': '391.5', 'completions/min_length': '331.8', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9625', 'completions/mean_terminated_length': '51.8', 'completions/min_terminated_length': '51.8', 'completions/max_terminated_length': '51.8', 'rewards/reward_fn/mean': '-0.017', 'rewards/reward_fn/std': '0.7561', 'reward': '-0.017', 'reward_std': '0.7561', 'frac_reward_zero_std': '0.975', 'entropy': '0.8178', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.51', 'epoch': '4.333'}
{'loss': '-0.01235', 'grad_norm': '0', 'learning_rate': '8.88e-06', 'num_tokens': '8.924e+05', 'completions/mean_length': '380.8', 'completions/min_length': '246.2', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9375', 'completions/mean_terminated_length': '46.2', 'completions/min_terminated_length': '46.2', 'completions/max_terminated_length': '46.2', 'rewards/reward_fn/mean': '-0.1348', 'rewards/reward_fn/std': '0.9628', 'reward': '-0.1348', 'reward_std': '0.9628', 'frac_reward_zero_std': '0.95', 'entropy': '0.8855', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.51', 'epoch': '4.667'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '8.08e-06', 'num_tokens': '9.569e+05', 'completions/mean_length': '394.3', 'completions/min_length': '361.9', 'completions/max_length': '400', 'completions/clipped_ratio': '0.975', 'completions/mean_terminated_length': '17.1', 'completions/min_terminated_length': '1.9', 'completions/max_terminated_length': '32.3', 'rewards/reward_fn/mean': '-0.03', 'rewards/reward_fn/std': '0.7298', 'reward': '-0.03', 'reward_std': '0.7298', 'frac_reward_zero_std': '1', 'entropy': '0.8237', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.47', 'epoch': '5'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '7.28e-06', 'num_tokens': '1.021e+06', 'completions/mean_length': '394.8', 'completions/min_length': '358.5', 'completions/max_length': '400', 'completions/clipped_ratio': '0.975', 'completions/mean_terminated_length': '38.5', 'completions/min_terminated_length': '38.5', 'completions/max_terminated_length': '38.5', 'rewards/reward_fn/mean': '-0.0625', 'rewards/reward_fn/std': '0.9605', 'reward': '-0.0625', 'reward_std': '0.9605', 'frac_reward_zero_std': '1', 'entropy': '0.9815', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.41', 'epoch': '5.333'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '6.48e-06', 'num_tokens': '1.086e+06', 'completions/mean_length': '392.6', 'completions/min_length': '341.2', 'completions/max_length': '400', 'completions/clipped_ratio': '0.975', 'completions/mean_terminated_length': '21.2', 'completions/min_terminated_length': '21.2', 'completions/max_terminated_length': '21.2', 'rewards/reward_fn/mean': '-0.1513', 'rewards/reward_fn/std': '0.8553', 'reward': '-0.1513', 'reward_std': '0.8553', 'frac_reward_zero_std': '0.975', 'entropy': '0.907', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.48', 'epoch': '5.667'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '5.68e-06', 'num_tokens': '1.151e+06', 'completions/mean_length': '400', 'completions/min_length': '400', 'completions/max_length': '400', 'completions/clipped_ratio': '1', 'completions/mean_terminated_length': '0', 'completions/min_terminated_length': '0', 'completions/max_terminated_length': '0', 'rewards/reward_fn/mean': '0.082', 'rewards/reward_fn/std': '0.9346', 'reward': '0.082', 'reward_std': '0.9346', 'frac_reward_zero_std': '1', 'entropy': '0.9227', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.47', 'epoch': '6'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '4.88e-06', 'num_tokens': '1.215e+06', 'completions/mean_length': '394.4', 'completions/min_length': '355.5', 'completions/max_length': '400', 'completions/clipped_ratio': '0.975', 'completions/mean_terminated_length': '35.5', 'completions/min_terminated_length': '35.5', 'completions/max_terminated_length': '35.5', 'rewards/reward_fn/mean': '-0.08775', 'rewards/reward_fn/std': '0.8883', 'reward': '-0.08775', 'reward_std': '0.8883', 'frac_reward_zero_std': '0.975', 'entropy': '0.9981', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.42', 'epoch': '6.333'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '4.08e-06', 'num_tokens': '1.28e+06', 'completions/mean_length': '398.8', 'completions/min_length': '390.4', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9875', 'completions/mean_terminated_length': '30.4', 'completions/min_terminated_length': '30.4', 'completions/max_terminated_length': '30.4', 'rewards/reward_fn/mean': '-0.03525', 'rewards/reward_fn/std': '0.8935', 'reward': '-0.03525', 'reward_std': '0.8935', 'frac_reward_zero_std': '1', 'entropy': '0.979', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.42', 'epoch': '6.667'}
Writing model shards: 0%| | 0/1 [00:00<?, ?it/s]
Writing model shards: 100%|██████████| 1/1 [00:16<00:00, 16.63s/it]
Writing model shards: 100%|██████████| 1/1 [00:16<00:00, 16.63s/it]
80%|████████ | 201/250 [1:58:00<36:25, 44.60s/it]
81%|████████ | 202/250 [1:58:36<33:26, 41.81s/it]
81%|████████ | 203/250 [1:59:11<31:07, 39.73s/it]
82%|████████▏ | 204/250 [1:59:46<29:29, 38.46s/it]
82%|████████▏ | 205/250 [2:00:22<28:10, 37.56s/it]
82%|████████▏ | 206/250 [2:00:57<27:03, 36.91s/it]
83%|████████▎ | 207/250 [2:01:32<25:57, 36.21s/it]
83%|████████▎ | 208/250 [2:02:06<25:00, 35.72s/it]
84%|████████▎ | 209/250 [2:02:41<24:15, 35.50s/it]
84%|████████▍ | 210/250 [2:03:16<23:37, 35.43s/it]
84%|████████▍ | 210/250 [2:03:16<23:37, 35.43s/it]
84%|████████▍ | 211/250 [2:03:52<22:59, 35.38s/it]
85%|████████▍ | 212/250 [2:04:26<22:17, 35.20s/it]
85%|████████▌ | 213/250 [2:05:01<21:38, 35.09s/it]
86%|████████▌ | 214/250 [2:05:36<20:55, 34.87s/it]
86%|████████▌ | 215/250 [2:06:11<20:24, 34.98s/it]
86%|████████▋ | 216/250 [2:06:46<19:49, 34.99s/it]
87%|████████▋ | 217/250 [2:07:21<19:11, 34.91s/it]
87%|████████▋ | 218/250 [2:07:55<18:36, 34.90s/it]
88%|████████▊ | 219/250 [2:08:30<18:01, 34.90s/it]
88%|████████▊ | 220/250 [2:09:05<17:27, 34.91s/it]
88%|████████▊ | 220/250 [2:09:05<17:27, 34.91s/it]
88%|████████▊ | 221/250 [2:09:40<16:52, 34.90s/it]
89%|████████▉ | 222/250 [2:10:16<16:21, 35.06s/it]
89%|████████▉ | 223/250 [2:10:50<15:44, 34.96s/it]
90%|████████▉ | 224/250 [2:11:25<15:09, 34.97s/it]
90%|█████████ | 225/250 [2:12:00<14:32, 34.92s/it]
90%|█████████ | 226/250 [2:12:35<13:57, 34.91s/it]
91%|█████████ | 227/250 [2:13:11<13:27, 35.12s/it]
91%|█████████ | 228/250 [2:13:45<12:49, 34.97s/it]
92%|█████████▏| 229/250 [2:14:21<12:17, 35.10s/it]
92%|█████████▏| 230/250 [2:14:55<11:39, 34.96s/it]
92%|█████████▏| 230/250 [2:14:55<11:39, 34.96s/it]
92%|█████████▏| 231/250 [2:15:31<11:06, 35.08s/it]
93%|█████████▎| 232/250 [2:16:06<10:32, 35.14s/it]
93%|█████████▎| 233/250 [2:16:42<10:00, 35.31s/it]
94%|█████████▎| 234/250 [2:17:17<09:24, 35.27s/it]
94%|█████████▍| 235/250 [2:17:52<08:47, 35.18s/it]
94%|█████████▍| 236/250 [2:18:26<08:09, 34.99s/it]
95%|█████████▍| 237/250 [2:19:01<07:33, 34.91s/it]
95%|█████████▌| 238/250 [2:19:36<06:58, 34.89s/it]
96%|█████████▌| 239/250 [2:20:12<06:26, 35.16s/it]
96%|█████████▌| 240/250 [2:20:47<05:52, 35.24s/it]
96%|█████████▌| 240/250 [2:20:47<05:52, 35.24s/it]
96%|█████████▋| 241/250 [2:21:22<05:16, 35.13s/it]
97%|█████████▋| 242/250 [2:21:57<04:40, 35.06s/it]
97%|█████████▋| 243/250 [2:22:32<04:05, 35.04s/it]
98%|█████████▊| 244/250 [2:23:07<03:30, 35.11s/it]
98%|█████████▊| 245/250 [2:23:42<02:55, 35.18s/it]
98%|█████████▊| 246/250 [2:24:18<02:20, 35.22s/it]
99%|█████████▉| 247/250 [2:24:53<01:45, 35.27s/it]
99%|█████████▉| 248/250 [2:25:28<01:10, 35.27s/it]
100%|█████████▉| 249/250 [2:26:03<00:35, 35.12s/it]
100%|██████████| 250/250 [2:26:38<00:00, 35.17s/it]
100%|██████████| 250/250 [2:26:39<00:00, 35.17s/it]{'loss': '0', 'grad_norm': '0', 'learning_rate': '3.28e-06', 'num_tokens': '1.344e+06', 'completions/mean_length': '392.1', 'completions/min_length': '337.2', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9625', 'completions/mean_terminated_length': '57.2', 'completions/min_terminated_length': '57.2', 'completions/max_terminated_length': '57.2', 'rewards/reward_fn/mean': '-0.0115', 'rewards/reward_fn/std': '0.8286', 'reward': '-0.0115', 'reward_std': '0.8286', 'frac_reward_zero_std': '1', 'entropy': '0.913', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.59', 'epoch': '7'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '2.48e-06', 'num_tokens': '1.409e+06', 'completions/mean_length': '400', 'completions/min_length': '400', 'completions/max_length': '400', 'completions/clipped_ratio': '1', 'completions/mean_terminated_length': '0', 'completions/min_terminated_length': '0', 'completions/max_terminated_length': '0', 'rewards/reward_fn/mean': '0.01825', 'rewards/reward_fn/std': '0.8384', 'reward': '0.01825', 'reward_std': '0.8384', 'frac_reward_zero_std': '1', 'entropy': '0.9604', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.32', 'epoch': '7.333'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '1.68e-06', 'num_tokens': '1.474e+06', 'completions/mean_length': '397.4', 'completions/min_length': '379.6', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9875', 'completions/mean_terminated_length': '19.6', 'completions/min_terminated_length': '19.6', 'completions/max_terminated_length': '19.6', 'rewards/reward_fn/mean': '-0.1123', 'rewards/reward_fn/std': '0.8594', 'reward': '-0.1123', 'reward_std': '0.8594', 'frac_reward_zero_std': '1', 'entropy': '0.9672', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.42', 'epoch': '7.667'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '8.8e-07', 'num_tokens': '1.538e+06', 'completions/mean_length': '389.5', 'completions/min_length': '316.1', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9625', 'completions/mean_terminated_length': '36.1', 'completions/min_terminated_length': '36.1', 'completions/max_terminated_length': '36.1', 'rewards/reward_fn/mean': '-0.015', 'rewards/reward_fn/std': '0.9199', 'reward': '-0.015', 'reward_std': '0.9199', 'frac_reward_zero_std': '1', 'entropy': '1.013', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.59', 'epoch': '8'}
{'loss': '0', 'grad_norm': '0', 'learning_rate': '8e-08', 'num_tokens': '1.603e+06', 'completions/mean_length': '391.8', 'completions/min_length': '334.6', 'completions/max_length': '400', 'completions/clipped_ratio': '0.9625', 'completions/mean_terminated_length': '54.6', 'completions/min_terminated_length': '54.6', 'completions/max_terminated_length': '54.6', 'rewards/reward_fn/mean': '0.0445', 'rewards/reward_fn/std': '0.861', 'reward': '0.0445', 'reward_std': '0.861', 'frac_reward_zero_std': '1', 'entropy': '0.9223', 'clip_ratio/low_mean': '0', 'clip_ratio/low_min': '0', 'clip_ratio/high_mean': '0', 'clip_ratio/high_max': '0', 'clip_ratio/region_mean': '0', 'step_time': '33.55', 'epoch': '8.333'}
Writing model shards: 0%| | 0/1 [00:00<?, ?it/s]
Writing model shards: 100%|██████████| 1/1 [00:05<00:00, 5.67s/it]
Writing model shards: 100%|██████████| 1/1 [00:05<00:00, 5.67s/it]
100%|██████████| 250/250 [2:26:57<00:00, 35.17s/it]
100%|██████████| 250/250 [2:26:57<00:00, 35.27s/it]
{'train_runtime': '8818', 'train_samples_per_second': '0.227', 'train_steps_per_second': '0.028', 'train_loss': '-0.001472', 'epoch': '8.333'}
[7] Saving model to ./deepsentinel_model_small
Writing model shards: 0%| | 0/1 [00:00<?, ?it/s]
Writing model shards: 100%|██████████| 1/1 [00:07<00:00, 7.07s/it]
Writing model shards: 100%|██████████| 1/1 [00:07<00:00, 7.08s/it]
Done!
[8] Post-training evaluation...
Baseline vs Post-training (heuristic used for comparison):
easy: -0.3681 → -0.3681 (Δ=+0.0000)
medium: 0.0049 → 0.0049 (Δ=+0.0000)
hard: 0.0833 → 0.0833 (Δ=+0.0000)
Training complete!
Model saved to: ./deepsentinel_model_small
Reference in New Issue View Git Blame Copy Permalink