logs/sft-v3.log

Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
Loading tokenizer + model: Qwen/Qwen3-8B

Loading weights:   0%|          | 0/399 [00:00<?, ?it/s]
Loading weights: 100%|██████████| 399/399 [00:00<00:00, 6506.82it/s]
[transformers] 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}.
trainable params: 30,670,848 || all params: 8,221,406,208 || trainable%: 0.3731
Building dataset from ../outputs/sft_data_5k_cot.jsonl (limit=None)
  3908 examples
Starting training

  0%|          | 0/490 [00:00<?, ?it/s]
  0%|          | 1/490 [00:01<08:35,  1.05s/it]
  0%|          | 2/490 [00:01<07:24,  1.10it/s]
  1%|          | 3/490 [00:02<06:53,  1.18it/s]
  1%|          | 4/490 [00:03<06:29,  1.25it/s]
  1%|          | 5/490 [00:04<06:21,  1.27it/s]
  1%|          | 6/490 [00:04<06:14,  1.29it/s]
  1%|▏         | 7/490 [00:05<06:01,  1.34it/s]
  2%|▏         | 8/490 [00:06<05:54,  1.36it/s]
  2%|▏         | 9/490 [00:06<05:44,  1.40it/s]
  2%|▏         | 10/490 [00:07<05:57,  1.34it/s]
                                                

  2%|▏         | 10/490 [00:07<05:57,  1.34it/s]
  2%|▏         | 11/490 [00:08<05:42,  1.40it/s]
  2%|▏         | 12/490 [00:09<05:38,  1.41it/s]
  3%|▎         | 13/490 [00:09<05:43,  1.39it/s]
  3%|▎         | 14/490 [00:10<05:43,  1.38it/s]
  3%|▎         | 15/490 [00:11<05:54,  1.34it/s]
  3%|▎         | 16/490 [00:12<05:54,  1.34it/s]
  3%|▎         | 17/490 [00:12<05:53,  1.34it/s]
  4%|▎         | 18/490 [00:13<05:45,  1.37it/s]
  4%|▍         | 19/490 [00:14<05:46,  1.36it/s]
  4%|▍         | 20/490 [00:15<05:44,  1.36it/s]
                                                

  4%|▍         | 20/490 [00:15<05:44,  1.36it/s]
  4%|▍         | 21/490 [00:15<05:44,  1.36it/s]
  4%|▍         | 22/490 [00:16<05:37,  1.39it/s]
  5%|▍         | 23/490 [00:17<05:34,  1.40it/s]
  5%|▍         | 24/490 [00:17<05:34,  1.39it/s]
  5%|▌         | 25/490 [00:18<05:38,  1.37it/s]
  5%|▌         | 26/490 [00:19<05:37,  1.37it/s]
  6%|▌         | 27/490 [00:20<05:28,  1.41it/s]
  6%|▌         | 28/490 [00:20<05:31,  1.39it/s]
  6%|▌         | 29/490 [00:21<05:32,  1.39it/s]
  6%|▌         | 30/490 [00:22<05:36,  1.37it/s]
                                                

  6%|▌         | 30/490 [00:22<05:36,  1.37it/s]
  6%|▋         | 31/490 [00:23<05:38,  1.36it/s]
  7%|▋         | 32/490 [00:23<05:37,  1.36it/s]
  7%|▋         | 33/490 [00:24<05:43,  1.33it/s]
  7%|▋         | 34/490 [00:25<05:33,  1.37it/s]
  7%|▋         | 35/490 [00:25<05:33,  1.37it/s]
  7%|▋         | 36/490 [00:26<05:26,  1.39it/s]
  8%|▊         | 37/490 [00:27<05:21,  1.41it/s]
  8%|▊         | 38/490 [00:28<05:19,  1.42it/s]
  8%|▊         | 39/490 [00:28<05:30,  1.36it/s]
  8%|▊         | 40/490 [00:29<05:30,  1.36it/s]
                                                

  8%|▊         | 40/490 [00:29<05:30,  1.36it/s]
  8%|▊         | 41/490 [00:30<05:34,  1.34it/s]
  9%|▊         | 42/490 [00:31<05:28,  1.37it/s]
  9%|▉         | 43/490 [00:31<05:29,  1.36it/s]
  9%|▉         | 44/490 [00:32<05:15,  1.41it/s]
  9%|▉         | 45/490 [00:33<05:22,  1.38it/s]
  9%|▉         | 46/490 [00:33<05:19,  1.39it/s]
 10%|▉         | 47/490 [00:34<05:19,  1.39it/s]
 10%|▉         | 48/490 [00:35<05:25,  1.36it/s]
 10%|█         | 49/490 [00:36<05:18,  1.38it/s]
 10%|█         | 50/490 [00:36<05:16,  1.39it/s]
                                                

 10%|█         | 50/490 [00:36<05:16,  1.39it/s]
 10%|█         | 51/490 [00:37<05:20,  1.37it/s]
 11%|█         | 52/490 [00:38<05:15,  1.39it/s]
 11%|█         | 53/490 [00:38<05:13,  1.40it/s]
 11%|█         | 54/490 [00:39<05:07,  1.42it/s]
 11%|█         | 55/490 [00:40<05:09,  1.40it/s]
 11%|█▏        | 56/490 [00:41<05:11,  1.39it/s]
 12%|█▏        | 57/490 [00:41<05:10,  1.39it/s]
 12%|█▏        | 58/490 [00:42<05:10,  1.39it/s]
 12%|█▏        | 59/490 [00:43<05:12,  1.38it/s]
 12%|█▏        | 60/490 [00:44<05:12,  1.37it/s]
                                                

 12%|█▏        | 60/490 [00:44<05:12,  1.37it/s]
 12%|█▏        | 61/490 [00:44<05:16,  1.36it/s]
 13%|█▎        | 62/490 [00:45<05:13,  1.36it/s]
 13%|█▎        | 63/490 [00:46<05:09,  1.38it/s]
 13%|█▎        | 64/490 [00:46<05:08,  1.38it/s]
 13%|█▎        | 65/490 [00:47<05:09,  1.37it/s]
 13%|█▎        | 66/490 [00:48<05:08,  1.37it/s]
 14%|█▎        | 67/490 [00:49<05:13,  1.35it/s]
 14%|█▍        | 68/490 [00:49<05:05,  1.38it/s]
 14%|█▍        | 69/490 [00:50<05:07,  1.37it/s]
 14%|█▍        | 70/490 [00:51<05:11,  1.35it/s]
                                                

 14%|█▍        | 70/490 [00:51<05:11,  1.35it/s]
 14%|█▍        | 71/490 [00:52<05:09,  1.36it/s]
 15%|█▍        | 72/490 [00:52<05:03,  1.38it/s]
 15%|█▍        | 73/490 [00:53<04:59,  1.39it/s]
 15%|█▌        | 74/490 [00:54<05:03,  1.37it/s]
 15%|█▌        | 75/490 [00:54<04:55,  1.40it/s]
 16%|█▌        | 76/490 [00:55<04:56,  1.40it/s]
 16%|█▌        | 77/490 [00:56<04:58,  1.38it/s]
 16%|█▌        | 78/490 [00:57<04:57,  1.39it/s]
 16%|█▌        | 79/490 [00:57<04:57,  1.38it/s]
 16%|█▋        | 80/490 [00:58<05:07,  1.34it/s]
                                                

 16%|█▋        | 80/490 [00:58<05:07,  1.34it/s]
 17%|█▋        | 81/490 [00:59<04:58,  1.37it/s]
 17%|█▋        | 82/490 [01:00<04:59,  1.36it/s]
 17%|█▋        | 83/490 [01:00<04:51,  1.40it/s]
 17%|█▋        | 84/490 [01:01<04:49,  1.40it/s]
 17%|█▋        | 85/490 [01:02<04:50,  1.39it/s]
 18%|█▊        | 86/490 [01:02<04:56,  1.36it/s]
 18%|█▊        | 87/490 [01:03<04:52,  1.38it/s]
 18%|█▊        | 88/490 [01:04<04:57,  1.35it/s]
 18%|█▊        | 89/490 [01:05<04:56,  1.35it/s]
 18%|█▊        | 90/490 [01:05<05:06,  1.30it/s]
                                                

 18%|█▊        | 90/490 [01:05<05:06,  1.30it/s]
 19%|█▊        | 91/490 [01:06<04:53,  1.36it/s]
 19%|█▉        | 92/490 [01:07<04:50,  1.37it/s]
 19%|█▉        | 93/490 [01:08<04:50,  1.36it/s]
 19%|█▉        | 94/490 [01:08<04:49,  1.37it/s]
 19%|█▉        | 95/490 [01:09<04:49,  1.36it/s]
 20%|█▉        | 96/490 [01:10<04:43,  1.39it/s]
 20%|█▉        | 97/490 [01:10<04:40,  1.40it/s]
 20%|██        | 98/490 [01:11<04:42,  1.39it/s]
 20%|██        | 99/490 [01:12<04:40,  1.39it/s]
 20%|██        | 100/490 [01:13<04:45,  1.36it/s]
                                                 

 20%|██        | 100/490 [01:13<04:45,  1.36it/s]
 21%|██        | 101/490 [01:13<04:46,  1.36it/s]
 21%|██        | 102/490 [01:14<04:47,  1.35it/s]
 21%|██        | 103/490 [01:15<04:41,  1.37it/s]
 21%|██        | 104/490 [01:16<04:36,  1.40it/s]
 21%|██▏       | 105/490 [01:16<04:36,  1.39it/s]
 22%|██▏       | 106/490 [01:17<04:35,  1.39it/s]
 22%|██▏       | 107/490 [01:18<04:32,  1.41it/s]
 22%|██▏       | 108/490 [01:18<04:27,  1.43it/s]
 22%|██▏       | 109/490 [01:19<04:27,  1.42it/s]
 22%|██▏       | 110/490 [01:20<04:29,  1.41it/s]
                                                 

 22%|██▏       | 110/490 [01:20<04:29,  1.41it/s]
 23%|██▎       | 111/490 [01:21<04:37,  1.37it/s]
 23%|██▎       | 112/490 [01:21<04:40,  1.35it/s]
 23%|██▎       | 113/490 [01:22<04:33,  1.38it/s]
 23%|██▎       | 114/490 [01:23<04:32,  1.38it/s]
 23%|██▎       | 115/490 [01:24<04:35,  1.36it/s]
 24%|██▎       | 116/490 [01:24<04:31,  1.38it/s]
 24%|██▍       | 117/490 [01:25<04:35,  1.35it/s]
 24%|██▍       | 118/490 [01:26<04:37,  1.34it/s]
 24%|██▍       | 119/490 [01:26<04:32,  1.36it/s]
 24%|██▍       | 120/490 [01:27<04:23,  1.40it/s]
                                                 

 24%|██▍       | 120/490 [01:27<04:23,  1.40it/s]
 25%|██▍       | 121/490 [01:28<04:21,  1.41it/s]
 25%|██▍       | 122/490 [01:29<04:17,  1.43it/s]
 25%|██▌       | 123/490 [01:29<04:21,  1.40it/s]
 25%|██▌       | 124/490 [01:30<04:21,  1.40it/s]
 26%|██▌       | 125/490 [01:31<04:31,  1.35it/s]
 26%|██▌       | 126/490 [01:31<04:26,  1.37it/s]
 26%|██▌       | 127/490 [01:32<04:19,  1.40it/s]
 26%|██▌       | 128/490 [01:33<04:14,  1.42it/s]
 26%|██▋       | 129/490 [01:34<04:09,  1.45it/s]
 27%|██▋       | 130/490 [01:34<04:13,  1.42it/s]
                                                 

 27%|██▋       | 130/490 [01:34<04:13,  1.42it/s]
 27%|██▋       | 131/490 [01:35<04:15,  1.40it/s]
 27%|██▋       | 132/490 [01:36<04:15,  1.40it/s]
 27%|██▋       | 133/490 [01:36<04:14,  1.40it/s]
 27%|██▋       | 134/490 [01:37<04:12,  1.41it/s]
 28%|██▊       | 135/490 [01:38<04:20,  1.36it/s]
 28%|██▊       | 136/490 [01:39<04:17,  1.37it/s]
 28%|██▊       | 137/490 [01:39<04:17,  1.37it/s]
 28%|██▊       | 138/490 [01:40<04:16,  1.37it/s]
 28%|██▊       | 139/490 [01:41<04:16,  1.37it/s]
 29%|██▊       | 140/490 [01:42<04:16,  1.36it/s]
                                                 

 29%|██▊       | 140/490 [01:42<04:16,  1.36it/s]
 29%|██▉       | 141/490 [01:42<04:11,  1.39it/s]
 29%|██▉       | 142/490 [01:43<04:10,  1.39it/s]
 29%|██▉       | 143/490 [01:44<04:08,  1.40it/s]
 29%|██▉       | 144/490 [01:44<04:05,  1.41it/s]
 30%|██▉       | 145/490 [01:45<04:08,  1.39it/s]
 30%|██▉       | 146/490 [01:46<04:08,  1.38it/s]
 30%|███       | 147/490 [01:47<04:10,  1.37it/s]
 30%|███       | 148/490 [01:47<04:10,  1.37it/s]
 30%|███       | 149/490 [01:48<04:15,  1.34it/s]
 31%|███       | 150/490 [01:49<04:07,  1.38it/s]
                                                 

 31%|███       | 150/490 [01:49<04:07,  1.38it/s]
 31%|███       | 151/490 [01:49<04:06,  1.37it/s]
 31%|███       | 152/490 [01:50<04:08,  1.36it/s]
 31%|███       | 153/490 [01:51<04:05,  1.37it/s]
 31%|███▏      | 154/490 [01:52<04:07,  1.36it/s]
 32%|███▏      | 155/490 [01:52<04:04,  1.37it/s]
 32%|███▏      | 156/490 [01:53<04:04,  1.36it/s]
 32%|███▏      | 157/490 [01:54<04:05,  1.36it/s]
 32%|███▏      | 158/490 [01:55<04:08,  1.34it/s]
 32%|███▏      | 159/490 [01:55<04:03,  1.36it/s]
 33%|███▎      | 160/490 [01:56<04:02,  1.36it/s]
                                                 

 33%|███▎      | 160/490 [01:56<04:02,  1.36it/s]
 33%|███▎      | 161/490 [01:57<04:00,  1.37it/s]
 33%|███▎      | 162/490 [01:58<03:54,  1.40it/s]
 33%|███▎      | 163/490 [01:58<03:55,  1.39it/s]
 33%|███▎      | 164/490 [01:59<03:57,  1.37it/s]
 34%|███▎      | 165/490 [02:00<03:53,  1.39it/s]
 34%|███▍      | 166/490 [02:00<03:51,  1.40it/s]
 34%|███▍      | 167/490 [02:01<03:49,  1.41it/s]
 34%|███▍      | 168/490 [02:02<03:45,  1.43it/s]
 34%|███▍      | 169/490 [02:03<03:50,  1.39it/s]
 35%|███▍      | 170/490 [02:03<03:49,  1.39it/s]
                                                 

 35%|███▍      | 170/490 [02:03<03:49,  1.39it/s]
 35%|███▍      | 171/490 [02:04<03:44,  1.42it/s]
 35%|███▌      | 172/490 [02:05<03:44,  1.42it/s]
 35%|███▌      | 173/490 [02:05<03:44,  1.41it/s]
 36%|███▌      | 174/490 [02:06<03:44,  1.41it/s]
 36%|███▌      | 175/490 [02:07<03:45,  1.40it/s]
 36%|███▌      | 176/490 [02:08<03:42,  1.41it/s]
 36%|███▌      | 177/490 [02:08<03:36,  1.44it/s]
 36%|███▋      | 178/490 [02:09<03:41,  1.41it/s]
 37%|███▋      | 179/490 [02:10<03:42,  1.40it/s]
 37%|███▋      | 180/490 [02:10<03:43,  1.39it/s]
                                                 

 37%|███▋      | 180/490 [02:10<03:43,  1.39it/s]
 37%|███▋      | 181/490 [02:11<03:46,  1.37it/s]
 37%|███▋      | 182/490 [02:12<03:45,  1.36it/s]
 37%|███▋      | 183/490 [02:13<03:41,  1.39it/s]
 38%|███▊      | 184/490 [02:13<03:38,  1.40it/s]
 38%|███▊      | 185/490 [02:14<03:43,  1.36it/s]
 38%|███▊      | 186/490 [02:15<03:46,  1.34it/s]
 38%|███▊      | 187/490 [02:16<03:48,  1.33it/s]
 38%|███▊      | 188/490 [02:16<03:40,  1.37it/s]
 39%|███▊      | 189/490 [02:17<03:36,  1.39it/s]
 39%|███▉      | 190/490 [02:18<03:37,  1.38it/s]
                                                 

 39%|███▉      | 190/490 [02:18<03:37,  1.38it/s]
 39%|███▉      | 191/490 [02:18<03:31,  1.42it/s]
 39%|███▉      | 192/490 [02:19<03:32,  1.40it/s]
 39%|███▉      | 193/490 [02:20<03:32,  1.39it/s]
 40%|███▉      | 194/490 [02:20<03:29,  1.42it/s]
 40%|███▉      | 195/490 [02:21<03:29,  1.41it/s]
 40%|████      | 196/490 [02:22<03:31,  1.39it/s]
 40%|████      | 197/490 [02:23<03:32,  1.38it/s]
 40%|████      | 198/490 [02:23<03:31,  1.38it/s]
 41%|████      | 199/490 [02:24<03:34,  1.36it/s]
 41%|████      | 200/490 [02:25<03:32,  1.37it/s]
                                                 

 41%|████      | 200/490 [02:25<03:32,  1.37it/s]
 41%|████      | 201/490 [02:26<04:10,  1.15it/s]
 41%|████      | 202/490 [02:27<03:54,  1.23it/s]
 41%|████▏     | 203/490 [02:27<03:42,  1.29it/s]
 42%|████▏     | 204/490 [02:28<03:36,  1.32it/s]
 42%|████▏     | 205/490 [02:29<03:30,  1.35it/s]
 42%|████▏     | 206/490 [02:30<03:36,  1.31it/s]
 42%|████▏     | 207/490 [02:30<03:33,  1.32it/s]
 42%|████▏     | 208/490 [02:31<03:34,  1.31it/s]
 43%|████▎     | 209/490 [02:32<03:33,  1.32it/s]
 43%|████▎     | 210/490 [02:33<03:34,  1.31it/s]
                                                 

 43%|████▎     | 210/490 [02:33<03:34,  1.31it/s]
 43%|████▎     | 211/490 [02:33<03:25,  1.36it/s]
 43%|████▎     | 212/490 [02:34<03:22,  1.38it/s]
 43%|████▎     | 213/490 [02:35<03:24,  1.35it/s]
 44%|████▎     | 214/490 [02:36<03:19,  1.38it/s]
 44%|████▍     | 215/490 [02:36<03:21,  1.37it/s]
 44%|████▍     | 216/490 [02:37<03:30,  1.30it/s]
 44%|████▍     | 217/490 [02:38<03:26,  1.32it/s]
 44%|████▍     | 218/490 [02:39<03:25,  1.32it/s]
 45%|████▍     | 219/490 [02:39<03:23,  1.33it/s]
 45%|████▍     | 220/490 [02:40<03:24,  1.32it/s]
                                                 

 45%|████▍     | 220/490 [02:40<03:24,  1.32it/s]
 45%|████▌     | 221/490 [02:41<03:22,  1.33it/s]
 45%|████▌     | 222/490 [02:42<03:21,  1.33it/s]
 46%|████▌     | 223/490 [02:42<03:15,  1.37it/s]
 46%|████▌     | 224/490 [02:43<03:12,  1.38it/s]
 46%|████▌     | 225/490 [02:44<03:09,  1.40it/s]
 46%|████▌     | 226/490 [02:44<03:12,  1.37it/s]
 46%|████▋     | 227/490 [02:45<03:13,  1.36it/s]
 47%|████▋     | 228/490 [02:46<03:10,  1.38it/s]
 47%|████▋     | 229/490 [02:47<03:04,  1.41it/s]
 47%|████▋     | 230/490 [02:47<03:00,  1.44it/s]
                                                 

 47%|████▋     | 230/490 [02:47<03:00,  1.44it/s]
 47%|████▋     | 231/490 [02:48<03:02,  1.42it/s]
 47%|████▋     | 232/490 [02:49<03:07,  1.38it/s]
 48%|████▊     | 233/490 [02:49<03:03,  1.40it/s]
 48%|████▊     | 234/490 [02:50<03:02,  1.40it/s]
 48%|████▊     | 235/490 [02:51<03:03,  1.39it/s]
 48%|████▊     | 236/490 [02:52<03:05,  1.37it/s]
 48%|████▊     | 237/490 [02:52<03:01,  1.39it/s]
 49%|████▊     | 238/490 [02:53<03:05,  1.36it/s]
 49%|████▉     | 239/490 [02:54<02:59,  1.40it/s]
 49%|████▉     | 240/490 [02:55<03:00,  1.39it/s]
                                                 

 49%|████▉     | 240/490 [02:55<03:00,  1.39it/s]
 49%|████▉     | 241/490 [02:55<02:56,  1.41it/s]
 49%|████▉     | 242/490 [02:56<02:59,  1.38it/s]
 50%|████▉     | 243/490 [02:57<03:01,  1.36it/s]
 50%|████▉     | 244/490 [02:57<02:56,  1.39it/s]
 50%|█████     | 245/490 [02:58<02:21,  1.73it/s]
 50%|█████     | 246/490 [02:58<02:31,  1.61it/s]
 50%|█████     | 247/490 [02:59<02:35,  1.57it/s]
 51%|█████     | 248/490 [03:00<02:43,  1.48it/s]
 51%|█████     | 249/490 [03:01<02:46,  1.45it/s]
 51%|█████     | 250/490 [03:01<02:44,  1.46it/s]
                                                 

 51%|█████     | 250/490 [03:01<02:44,  1.46it/s]
 51%|█████     | 251/490 [03:02<02:44,  1.45it/s]
 51%|█████▏    | 252/490 [03:03<02:47,  1.42it/s]
 52%|█████▏    | 253/490 [03:03<02:45,  1.43it/s]
 52%|█████▏    | 254/490 [03:04<02:45,  1.42it/s]
 52%|█████▏    | 255/490 [03:05<02:44,  1.43it/s]
 52%|█████▏    | 256/490 [03:05<02:42,  1.44it/s]
 52%|█████▏    | 257/490 [03:06<02:41,  1.44it/s]
 53%|█████▎    | 258/490 [03:07<02:41,  1.43it/s]
 53%|█████▎    | 259/490 [03:08<02:44,  1.40it/s]
 53%|█████▎    | 260/490 [03:08<02:41,  1.43it/s]
                                                 

 53%|█████▎    | 260/490 [03:08<02:41,  1.43it/s]
 53%|█████▎    | 261/490 [03:09<02:40,  1.42it/s]
 53%|█████▎    | 262/490 [03:10<02:42,  1.40it/s]
 54%|█████▎    | 263/490 [03:10<02:44,  1.38it/s]
 54%|█████▍    | 264/490 [03:11<02:44,  1.37it/s]
 54%|█████▍    | 265/490 [03:12<02:46,  1.36it/s]
 54%|█████▍    | 266/490 [03:13<02:44,  1.36it/s]
 54%|█████▍    | 267/490 [03:13<02:45,  1.35it/s]
 55%|█████▍    | 268/490 [03:14<02:45,  1.34it/s]
 55%|█████▍    | 269/490 [03:15<02:44,  1.35it/s]
 55%|█████▌    | 270/490 [03:16<02:43,  1.35it/s]
                                                 

 55%|█████▌    | 270/490 [03:16<02:43,  1.35it/s]
 55%|█████▌    | 271/490 [03:16<02:41,  1.36it/s]
 56%|█████▌    | 272/490 [03:17<02:39,  1.37it/s]
 56%|█████▌    | 273/490 [03:18<02:36,  1.39it/s]
 56%|█████▌    | 274/490 [03:19<02:34,  1.40it/s]
 56%|█████▌    | 275/490 [03:19<02:36,  1.38it/s]
 56%|█████▋    | 276/490 [03:20<02:40,  1.34it/s]
 57%|█████▋    | 277/490 [03:21<02:40,  1.32it/s]
 57%|█████▋    | 278/490 [03:22<02:35,  1.37it/s]
 57%|█████▋    | 279/490 [03:22<02:32,  1.38it/s]
 57%|█████▋    | 280/490 [03:23<02:29,  1.40it/s]
                                                 

 57%|█████▋    | 280/490 [03:23<02:29,  1.40it/s]
 57%|█████▋    | 281/490 [03:24<02:29,  1.40it/s]
 58%|█████▊    | 282/490 [03:24<02:28,  1.40it/s]
 58%|█████▊    | 283/490 [03:25<02:27,  1.40it/s]
 58%|█████▊    | 284/490 [03:26<02:28,  1.39it/s]
 58%|█████▊    | 285/490 [03:26<02:22,  1.44it/s]
 58%|█████▊    | 286/490 [03:27<02:20,  1.45it/s]
 59%|█████▊    | 287/490 [03:28<02:25,  1.39it/s]
 59%|█████▉    | 288/490 [03:29<02:26,  1.38it/s]
 59%|█████▉    | 289/490 [03:29<02:28,  1.35it/s]
 59%|█████▉    | 290/490 [03:30<02:26,  1.36it/s]
                                                 

 59%|█████▉    | 290/490 [03:30<02:26,  1.36it/s]
 59%|█████▉    | 291/490 [03:31<02:24,  1.37it/s]
 60%|█████▉    | 292/490 [03:32<02:23,  1.38it/s]
 60%|█████▉    | 293/490 [03:32<02:18,  1.42it/s]
 60%|██████    | 294/490 [03:33<02:22,  1.37it/s]
 60%|██████    | 295/490 [03:34<02:19,  1.40it/s]
 60%|██████    | 296/490 [03:34<02:20,  1.39it/s]
 61%|██████    | 297/490 [03:35<02:22,  1.35it/s]
 61%|██████    | 298/490 [03:36<02:24,  1.33it/s]
 61%|██████    | 299/490 [03:37<02:22,  1.34it/s]
 61%|██████    | 300/490 [03:37<02:18,  1.38it/s]
                                                 

 61%|██████    | 300/490 [03:37<02:18,  1.38it/s]
 61%|██████▏   | 301/490 [03:38<02:15,  1.40it/s]
 62%|██████▏   | 302/490 [03:39<02:13,  1.41it/s]
 62%|██████▏   | 303/490 [03:40<02:14,  1.39it/s]
 62%|██████▏   | 304/490 [03:40<02:16,  1.37it/s]
 62%|██████▏   | 305/490 [03:41<02:17,  1.35it/s]
 62%|██████▏   | 306/490 [03:42<02:16,  1.35it/s]
 63%|██████▎   | 307/490 [03:42<02:14,  1.36it/s]
 63%|██████▎   | 308/490 [03:43<02:11,  1.38it/s]
 63%|██████▎   | 309/490 [03:44<02:12,  1.37it/s]
 63%|██████▎   | 310/490 [03:45<02:13,  1.34it/s]
                                                 

 63%|██████▎   | 310/490 [03:45<02:13,  1.34it/s]
 63%|██████▎   | 311/490 [03:45<02:10,  1.37it/s]
 64%|██████▎   | 312/490 [03:46<02:08,  1.39it/s]
 64%|██████▍   | 313/490 [03:47<02:08,  1.38it/s]
 64%|██████▍   | 314/490 [03:48<02:09,  1.35it/s]
 64%|██████▍   | 315/490 [03:48<02:06,  1.38it/s]
 64%|██████▍   | 316/490 [03:49<02:03,  1.41it/s]
 65%|██████▍   | 317/490 [03:50<02:02,  1.41it/s]
 65%|██████▍   | 318/490 [03:50<02:02,  1.41it/s]
 65%|██████▌   | 319/490 [03:51<02:04,  1.37it/s]
 65%|██████▌   | 320/490 [03:52<02:00,  1.41it/s]
                                                 

 65%|██████▌   | 320/490 [03:52<02:00,  1.41it/s]
 66%|██████▌   | 321/490 [03:53<01:58,  1.43it/s]
 66%|██████▌   | 322/490 [03:53<01:57,  1.43it/s]
 66%|██████▌   | 323/490 [03:54<02:01,  1.37it/s]
 66%|██████▌   | 324/490 [03:55<01:58,  1.40it/s]
 66%|██████▋   | 325/490 [03:56<02:03,  1.34it/s]
 67%|██████▋   | 326/490 [03:56<02:02,  1.34it/s]
 67%|██████▋   | 327/490 [03:57<01:59,  1.36it/s]
 67%|██████▋   | 328/490 [03:58<01:57,  1.38it/s]
 67%|██████▋   | 329/490 [03:58<01:55,  1.39it/s]
 67%|██████▋   | 330/490 [03:59<01:56,  1.38it/s]
                                                 

 67%|██████▋   | 330/490 [03:59<01:56,  1.38it/s]
 68%|██████▊   | 331/490 [04:00<01:54,  1.39it/s]
 68%|██████▊   | 332/490 [04:01<01:52,  1.40it/s]
 68%|██████▊   | 333/490 [04:01<01:56,  1.35it/s]
 68%|██████▊   | 334/490 [04:02<01:54,  1.36it/s]
 68%|██████▊   | 335/490 [04:03<01:52,  1.37it/s]
 69%|██████▊   | 336/490 [04:03<01:51,  1.39it/s]
 69%|██████▉   | 337/490 [04:04<01:49,  1.40it/s]
 69%|██████▉   | 338/490 [04:05<01:46,  1.42it/s]
 69%|██████▉   | 339/490 [04:06<01:49,  1.37it/s]
 69%|██████▉   | 340/490 [04:06<01:52,  1.34it/s]
                                                 

 69%|██████▉   | 340/490 [04:06<01:52,  1.34it/s]
 70%|██████▉   | 341/490 [04:07<01:51,  1.34it/s]
 70%|██████▉   | 342/490 [04:08<01:50,  1.34it/s]
 70%|███████   | 343/490 [04:09<01:50,  1.33it/s]
 70%|███████   | 344/490 [04:09<01:48,  1.35it/s]
 70%|███████   | 345/490 [04:10<01:45,  1.37it/s]
 71%|███████   | 346/490 [04:11<01:43,  1.40it/s]
 71%|███████   | 347/490 [04:12<01:46,  1.35it/s]
 71%|███████   | 348/490 [04:12<01:42,  1.39it/s]
 71%|███████   | 349/490 [04:13<01:41,  1.39it/s]
 71%|███████▏  | 350/490 [04:14<01:39,  1.40it/s]
                                                 

 71%|███████▏  | 350/490 [04:14<01:39,  1.40it/s]
 72%|███████▏  | 351/490 [04:14<01:41,  1.37it/s]
 72%|███████▏  | 352/490 [04:15<01:37,  1.41it/s]
 72%|███████▏  | 353/490 [04:16<01:35,  1.44it/s]
 72%|███████▏  | 354/490 [04:17<01:37,  1.40it/s]
 72%|███████▏  | 355/490 [04:17<01:36,  1.40it/s]
 73%|███████▎  | 356/490 [04:18<01:37,  1.38it/s]
 73%|███████▎  | 357/490 [04:19<01:35,  1.39it/s]
 73%|███████▎  | 358/490 [04:19<01:35,  1.38it/s]
 73%|███████▎  | 359/490 [04:20<01:34,  1.39it/s]
 73%|███████▎  | 360/490 [04:21<01:34,  1.38it/s]
                                                 

 73%|███████▎  | 360/490 [04:21<01:34,  1.38it/s]
 74%|███████▎  | 361/490 [04:22<01:33,  1.38it/s]
 74%|███████▍  | 362/490 [04:22<01:32,  1.38it/s]
 74%|███████▍  | 363/490 [04:23<01:31,  1.39it/s]
 74%|███████▍  | 364/490 [04:24<01:28,  1.42it/s]
 74%|███████▍  | 365/490 [04:24<01:29,  1.39it/s]
 75%|███████▍  | 366/490 [04:25<01:28,  1.40it/s]
 75%|███████▍  | 367/490 [04:26<01:27,  1.41it/s]
 75%|███████▌  | 368/490 [04:27<01:28,  1.37it/s]
 75%|███████▌  | 369/490 [04:27<01:29,  1.36it/s]
 76%|███████▌  | 370/490 [04:28<01:26,  1.38it/s]
                                                 

 76%|███████▌  | 370/490 [04:28<01:26,  1.38it/s]
 76%|███████▌  | 371/490 [04:29<01:27,  1.36it/s]
 76%|███████▌  | 372/490 [04:30<01:24,  1.39it/s]
 76%|███████▌  | 373/490 [04:30<01:27,  1.34it/s]
 76%|███████▋  | 374/490 [04:31<01:24,  1.37it/s]
 77%|███████▋  | 375/490 [04:32<01:22,  1.40it/s]
 77%|███████▋  | 376/490 [04:32<01:20,  1.41it/s]
 77%|███████▋  | 377/490 [04:33<01:19,  1.42it/s]
 77%|███████▋  | 378/490 [04:34<01:19,  1.42it/s]
 77%|███████▋  | 379/490 [04:35<01:18,  1.42it/s]
 78%|███████▊  | 380/490 [04:35<01:17,  1.41it/s]
                                                 

 78%|███████▊  | 380/490 [04:35<01:17,  1.41it/s]
 78%|███████▊  | 381/490 [04:36<01:21,  1.34it/s]
 78%|███████▊  | 382/490 [04:37<01:20,  1.35it/s]
 78%|███████▊  | 383/490 [04:38<01:19,  1.35it/s]
 78%|███████▊  | 384/490 [04:38<01:18,  1.36it/s]
 79%|███████▊  | 385/490 [04:39<01:16,  1.38it/s]
 79%|███████▉  | 386/490 [04:40<01:13,  1.42it/s]
 79%|███████▉  | 387/490 [04:40<01:13,  1.40it/s]
 79%|███████▉  | 388/490 [04:41<01:15,  1.36it/s]
 79%|███████▉  | 389/490 [04:42<01:15,  1.34it/s]
 80%|███████▉  | 390/490 [04:43<01:13,  1.37it/s]
                                                 

 80%|███████▉  | 390/490 [04:43<01:13,  1.37it/s]
 80%|███████▉  | 391/490 [04:43<01:13,  1.35it/s]
 80%|████████  | 392/490 [04:44<01:11,  1.37it/s]
 80%|████████  | 393/490 [04:45<01:11,  1.36it/s]
 80%|████████  | 394/490 [04:46<01:09,  1.38it/s]
 81%|████████  | 395/490 [04:46<01:09,  1.37it/s]
 81%|████████  | 396/490 [04:47<01:09,  1.36it/s]
 81%|████████  | 397/490 [04:48<01:07,  1.37it/s]
 81%|████████  | 398/490 [04:48<01:07,  1.36it/s]
 81%|████████▏ | 399/490 [04:49<01:05,  1.38it/s]
 82%|████████▏ | 400/490 [04:50<01:06,  1.36it/s]
                                                 

 82%|████████▏ | 400/490 [04:50<01:06,  1.36it/s]
 82%|████████▏ | 401/490 [04:51<01:15,  1.18it/s]
 82%|████████▏ | 402/490 [04:52<01:10,  1.26it/s]
 82%|████████▏ | 403/490 [04:52<01:05,  1.32it/s]
 82%|████████▏ | 404/490 [04:53<01:05,  1.32it/s]
 83%|████████▎ | 405/490 [04:54<01:02,  1.36it/s]
 83%|████████▎ | 406/490 [04:55<01:00,  1.38it/s]
 83%|████████▎ | 407/490 [04:55<01:00,  1.37it/s]
 83%|████████▎ | 408/490 [04:56<00:58,  1.39it/s]
 83%|████████▎ | 409/490 [04:57<00:58,  1.38it/s]
 84%|████████▎ | 410/490 [04:57<00:57,  1.38it/s]
                                                 

 84%|████████▎ | 410/490 [04:57<00:57,  1.38it/s]
 84%|████████▍ | 411/490 [04:58<00:57,  1.38it/s]
 84%|████████▍ | 412/490 [04:59<00:55,  1.39it/s]
 84%|████████▍ | 413/490 [05:00<00:57,  1.33it/s]
 84%|████████▍ | 414/490 [05:00<00:55,  1.37it/s]
 85%|████████▍ | 415/490 [05:01<00:54,  1.38it/s]
 85%|████████▍ | 416/490 [05:02<00:52,  1.40it/s]
 85%|████████▌ | 417/490 [05:02<00:51,  1.41it/s]
 85%|████████▌ | 418/490 [05:03<00:52,  1.38it/s]
 86%|████████▌ | 419/490 [05:04<00:51,  1.37it/s]
 86%|████████▌ | 420/490 [05:05<00:50,  1.38it/s]
                                                 

 86%|████████▌ | 420/490 [05:05<00:50,  1.38it/s]
 86%|████████▌ | 421/490 [05:05<00:49,  1.40it/s]
 86%|████████▌ | 422/490 [05:06<00:49,  1.39it/s]
 86%|████████▋ | 423/490 [05:07<00:47,  1.40it/s]
 87%|████████▋ | 424/490 [05:07<00:46,  1.41it/s]
 87%|████████▋ | 425/490 [05:08<00:47,  1.38it/s]
 87%|████████▋ | 426/490 [05:09<00:46,  1.36it/s]
 87%|████████▋ | 427/490 [05:10<00:46,  1.36it/s]
 87%|████████▋ | 428/490 [05:10<00:44,  1.38it/s]
 88%|████████▊ | 429/490 [05:11<00:43,  1.40it/s]
 88%|████████▊ | 430/490 [05:12<00:42,  1.41it/s]
                                                 

 88%|████████▊ | 430/490 [05:12<00:42,  1.41it/s]
 88%|████████▊ | 431/490 [05:13<00:41,  1.41it/s]
 88%|████████▊ | 432/490 [05:13<00:41,  1.40it/s]
 88%|████████▊ | 433/490 [05:14<00:43,  1.32it/s]
 89%|████████▊ | 434/490 [05:15<00:41,  1.35it/s]
 89%|████████▉ | 435/490 [05:16<00:40,  1.34it/s]
 89%|████████▉ | 436/490 [05:16<00:39,  1.36it/s]
 89%|████████▉ | 437/490 [05:17<00:38,  1.37it/s]
 89%|████████▉ | 438/490 [05:18<00:37,  1.38it/s]
 90%|████████▉ | 439/490 [05:18<00:37,  1.37it/s]
 90%|████████▉ | 440/490 [05:19<00:35,  1.39it/s]
                                                 

 90%|████████▉ | 440/490 [05:19<00:35,  1.39it/s]
 90%|█████████ | 441/490 [05:20<00:35,  1.38it/s]
 90%|█████████ | 442/490 [05:21<00:34,  1.39it/s]
 90%|█████████ | 443/490 [05:21<00:33,  1.41it/s]
 91%|█████████ | 444/490 [05:22<00:33,  1.37it/s]
 91%|█████████ | 445/490 [05:23<00:32,  1.39it/s]
 91%|█████████ | 446/490 [05:23<00:31,  1.40it/s]
 91%|█████████ | 447/490 [05:24<00:30,  1.40it/s]
 91%|█████████▏| 448/490 [05:25<00:30,  1.38it/s]
 92%|█████████▏| 449/490 [05:26<00:29,  1.40it/s]
 92%|█████████▏| 450/490 [05:26<00:28,  1.41it/s]
                                                 

 92%|█████████▏| 450/490 [05:26<00:28,  1.41it/s]
 92%|█████████▏| 451/490 [05:27<00:28,  1.38it/s]
 92%|█████████▏| 452/490 [05:28<00:27,  1.39it/s]
 92%|█████████▏| 453/490 [05:29<00:26,  1.39it/s]
 93%|█████████▎| 454/490 [05:29<00:25,  1.39it/s]
 93%|█████████▎| 455/490 [05:30<00:24,  1.40it/s]
 93%|█████████▎| 456/490 [05:31<00:24,  1.40it/s]
 93%|█████████▎| 457/490 [05:31<00:23,  1.39it/s]
 93%|█████████▎| 458/490 [05:32<00:22,  1.43it/s]
 94%|█████████▎| 459/490 [05:33<00:21,  1.42it/s]
 94%|█████████▍| 460/490 [05:33<00:20,  1.43it/s]
                                                 

 94%|█████████▍| 460/490 [05:33<00:20,  1.43it/s]
 94%|█████████▍| 461/490 [05:34<00:20,  1.41it/s]
 94%|█████████▍| 462/490 [05:35<00:19,  1.43it/s]
 94%|█████████▍| 463/490 [05:36<00:19,  1.40it/s]
 95%|█████████▍| 464/490 [05:36<00:18,  1.38it/s]
 95%|█████████▍| 465/490 [05:37<00:18,  1.35it/s]
 95%|█████████▌| 466/490 [05:38<00:17,  1.36it/s]
 95%|█████████▌| 467/490 [05:39<00:16,  1.36it/s]
 96%|█████████▌| 468/490 [05:39<00:16,  1.36it/s]
 96%|█████████▌| 469/490 [05:40<00:15,  1.37it/s]
 96%|█████████▌| 470/490 [05:41<00:14,  1.37it/s]
                                                 

 96%|█████████▌| 470/490 [05:41<00:14,  1.37it/s]
 96%|█████████▌| 471/490 [05:42<00:14,  1.35it/s]
 96%|█████████▋| 472/490 [05:42<00:13,  1.35it/s]
 97%|█████████▋| 473/490 [05:43<00:12,  1.33it/s]
 97%|█████████▋| 474/490 [05:44<00:12,  1.31it/s]
 97%|█████████▋| 475/490 [05:45<00:11,  1.35it/s]
 97%|█████████▋| 476/490 [05:45<00:10,  1.37it/s]
 97%|█████████▋| 477/490 [05:46<00:09,  1.37it/s]
 98%|█████████▊| 478/490 [05:47<00:08,  1.38it/s]
 98%|█████████▊| 479/490 [05:47<00:08,  1.37it/s]
 98%|█████████▊| 480/490 [05:48<00:07,  1.38it/s]
                                                 

 98%|█████████▊| 480/490 [05:48<00:07,  1.38it/s]
 98%|█████████▊| 481/490 [05:49<00:06,  1.36it/s]
 98%|█████████▊| 482/490 [05:50<00:05,  1.34it/s]
 99%|█████████▊| 483/490 [05:50<00:05,  1.37it/s]
 99%|█████████▉| 484/490 [05:51<00:04,  1.34it/s]
 99%|█████████▉| 485/490 [05:52<00:03,  1.31it/s]
 99%|█████████▉| 486/490 [05:53<00:02,  1.35it/s]
 99%|█████████▉| 487/490 [05:53<00:02,  1.31it/s]
100%|█████████▉| 488/490 [05:54<00:01,  1.34it/s]
100%|█████████▉| 489/490 [05:55<00:00,  1.36it/s]
100%|██████████| 490/490 [05:55<00:00,  1.69it/s]
                                                 

100%|██████████| 490/490 [05:55<00:00,  1.69it/s]
                                                 

100%|██████████| 490/490 [05:56<00:00,  1.69it/s]
100%|██████████| 490/490 [05:56<00:00,  1.38it/s]
{'loss': '1.959', 'grad_norm': '2.154', 'learning_rate': '9e-07', 'epoch': '0.0409'}
{'loss': '1.951', 'grad_norm': '2.518', 'learning_rate': '1.9e-06', 'epoch': '0.0818'}
{'loss': '1.836', 'grad_norm': '2.256', 'learning_rate': '2.9e-06', 'epoch': '0.1227'}
{'loss': '1.931', 'grad_norm': '2.525', 'learning_rate': '3.9e-06', 'epoch': '0.1636'}
{'loss': '1.817', 'grad_norm': '2.297', 'learning_rate': '4.9e-06', 'epoch': '0.2045'}
{'loss': '1.556', 'grad_norm': '1.899', 'learning_rate': '5.9e-06', 'epoch': '0.2454'}
{'loss': '1.333', 'grad_norm': '1.628', 'learning_rate': '6.9e-06', 'epoch': '0.2863'}
{'loss': '1.168', 'grad_norm': '1.474', 'learning_rate': '7.9e-06', 'epoch': '0.3272'}
{'loss': '0.8124', 'grad_norm': '1.223', 'learning_rate': '8.9e-06', 'epoch': '0.3681'}
{'loss': '0.5554', 'grad_norm': '1.163', 'learning_rate': '9.9e-06', 'epoch': '0.409'}
{'loss': '0.3553', 'grad_norm': '0.9885', 'learning_rate': '9.769e-06', 'epoch': '0.4499'}
{'loss': '0.1804', 'grad_norm': '0.7532', 'learning_rate': '9.513e-06', 'epoch': '0.4908'}
{'loss': '0.1795', 'grad_norm': '0.7922', 'learning_rate': '9.256e-06', 'epoch': '0.5317'}
{'loss': '0.1708', 'grad_norm': '0.5374', 'learning_rate': '9e-06', 'epoch': '0.5726'}
{'loss': '0.1618', 'grad_norm': '0.4581', 'learning_rate': '8.744e-06', 'epoch': '0.6135'}
{'loss': '0.1472', 'grad_norm': '0.9656', 'learning_rate': '8.487e-06', 'epoch': '0.6544'}
{'loss': '0.1498', 'grad_norm': '0.4566', 'learning_rate': '8.231e-06', 'epoch': '0.6953'}
{'loss': '0.124', 'grad_norm': '0.3963', 'learning_rate': '7.974e-06', 'epoch': '0.7362'}
{'loss': '0.1395', 'grad_norm': '1.108', 'learning_rate': '7.718e-06', 'epoch': '0.7771'}
{'loss': '0.1342', 'grad_norm': '0.6897', 'learning_rate': '7.462e-06', 'epoch': '0.818'}
{'loss': '0.1101', 'grad_norm': '0.4212', 'learning_rate': '7.205e-06', 'epoch': '0.8589'}
{'loss': '0.1009', 'grad_norm': '0.4124', 'learning_rate': '6.949e-06', 'epoch': '0.8998'}
{'loss': '0.1086', 'grad_norm': '0.4313', 'learning_rate': '6.692e-06', 'epoch': '0.9407'}
{'loss': '0.08618', 'grad_norm': '0.7137', 'learning_rate': '6.436e-06', 'epoch': '0.9816'}
{'loss': '0.1261', 'grad_norm': '0.2955', 'learning_rate': '6.179e-06', 'epoch': '1.02'}
{'loss': '0.09421', 'grad_norm': '0.6246', 'learning_rate': '5.923e-06', 'epoch': '1.061'}
{'loss': '0.09973', 'grad_norm': '0.4065', 'learning_rate': '5.667e-06', 'epoch': '1.102'}
{'loss': '0.1033', 'grad_norm': '0.4332', 'learning_rate': '5.41e-06', 'epoch': '1.143'}
{'loss': '0.08006', 'grad_norm': '0.3733', 'learning_rate': '5.154e-06', 'epoch': '1.184'}
{'loss': '0.1023', 'grad_norm': '0.4223', 'learning_rate': '4.897e-06', 'epoch': '1.225'}
{'loss': '0.1122', 'grad_norm': '0.5237', 'learning_rate': '4.641e-06', 'epoch': '1.266'}
{'loss': '0.1198', 'grad_norm': '0.5534', 'learning_rate': '4.385e-06', 'epoch': '1.307'}
{'loss': '0.09909', 'grad_norm': '0.6251', 'learning_rate': '4.128e-06', 'epoch': '1.348'}
{'loss': '0.1054', 'grad_norm': '0.4749', 'learning_rate': '3.872e-06', 'epoch': '1.389'}
{'loss': '0.08617', 'grad_norm': '0.344', 'learning_rate': '3.615e-06', 'epoch': '1.429'}
{'loss': '0.1053', 'grad_norm': '0.9175', 'learning_rate': '3.359e-06', 'epoch': '1.47'}
{'loss': '0.06682', 'grad_norm': '0.4685', 'learning_rate': '3.103e-06', 'epoch': '1.511'}
{'loss': '0.1269', 'grad_norm': '1.05', 'learning_rate': '2.846e-06', 'epoch': '1.552'}
{'loss': '0.09101', 'grad_norm': '0.3128', 'learning_rate': '2.59e-06', 'epoch': '1.593'}
{'loss': '0.09549', 'grad_norm': '0.48', 'learning_rate': '2.333e-06', 'epoch': '1.634'}
{'loss': '0.09072', 'grad_norm': '0.3007', 'learning_rate': '2.077e-06', 'epoch': '1.675'}
{'loss': '0.0969', 'grad_norm': '0.5814', 'learning_rate': '1.821e-06', 'epoch': '1.716'}
{'loss': '0.0776', 'grad_norm': '0.4144', 'learning_rate': '1.564e-06', 'epoch': '1.757'}
{'loss': '0.08835', 'grad_norm': '0.7711', 'learning_rate': '1.308e-06', 'epoch': '1.798'}
{'loss': '0.08903', 'grad_norm': '0.2714', 'learning_rate': '1.051e-06', 'epoch': '1.838'}
{'loss': '0.08643', 'grad_norm': '0.3585', 'learning_rate': '7.949e-07', 'epoch': '1.879'}
{'loss': '0.07803', 'grad_norm': '0.4993', 'learning_rate': '5.385e-07', 'epoch': '1.92'}
{'loss': '0.08828', 'grad_norm': '0.5227', 'learning_rate': '2.821e-07', 'epoch': '1.961'}
{'loss': '0.09387', 'grad_norm': '1.508', 'learning_rate': '2.564e-08', 'epoch': '2'}
{'train_runtime': '356', 'train_samples_per_second': '21.95', 'train_steps_per_second': '1.376', 'train_loss': '0.3973', 'epoch': '2'}
Saved final model to ../checkpoints/sft-v3/final