Loaded 100 samples from /mnt/hdfs/byte_tt_data_cu_vagcp/haogeng.liu/new_policy7w_v2_reformat
Loading model: /mnt/bn/bohanzhainas1/jiashuo/exp/new_policy7w_v2_reformat/checkpoint-1700/hf_model
`torch_dtype` is deprecated! Use `dtype` instead!

Loading checkpoint shards:   0%|          | 0/5 [00:00<?, ?it/s]
Loading checkpoint shards:  20%|██        | 1/5 [00:01<00:04,  1.18s/it]
Loading checkpoint shards:  40%|████      | 2/5 [00:02<00:03,  1.16s/it]
Loading checkpoint shards:  60%|██████    | 3/5 [00:03<00:02,  1.16s/it]
Loading checkpoint shards:  80%|████████  | 4/5 [00:03<00:00,  1.34it/s]
Loading checkpoint shards: 100%|██████████| 5/5 [00:03<00:00,  1.70it/s]
Loading checkpoint shards: 100%|██████████| 5/5 [00:03<00:00,  1.28it/s]
The tokenizer you are loading from '/mnt/bn/bohanzhainas1/jiashuo/exp/new_policy7w_v2_reformat/checkpoint-1700/hf_model' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.
Model loaded.
[1/100] gt=0 pred=0 match=True | acc=1.000 (1/1) | 0.0 samp/s
[2/100] gt=1 pred=1 match=True | acc=1.000 (2/2) | 0.1 samp/s
[3/100] gt=1 pred=1 match=True | acc=1.000 (3/3) | 0.1 samp/s
[4/100] gt=1 pred=1 match=True | acc=1.000 (4/4) | 0.0 samp/s
[5/100] gt=0 pred=0 match=True | acc=1.000 (5/5) | 0.1 samp/s
[6/100] gt=0 pred=0 match=True | acc=1.000 (6/6) | 0.1 samp/s
[7/100] gt=0 pred=0 match=True | acc=1.000 (7/7) | 0.1 samp/s
[8/100] gt=0 pred=0 match=True | acc=1.000 (8/8) | 0.1 samp/s
[9/100] gt=0 pred=0 match=True | acc=1.000 (9/9) | 0.1 samp/s
[10/100] gt=0 pred=0 match=True | acc=1.000 (10/10) | 0.1 samp/s
[11/100] gt=0 pred=0 match=True | acc=1.000 (11/11) | 0.1 samp/s
[12/100] gt=0 pred=0 match=True | acc=1.000 (12/12) | 0.1 samp/s
[13/100] gt=0 pred=0 match=True | acc=1.000 (13/13) | 0.1 samp/s
[14/100] gt=0 pred=0 match=True | acc=1.000 (14/14) | 0.1 samp/s
[15/100] gt=0 pred=1 match=False | acc=0.933 (14/15) | 0.1 samp/s
[16/100] gt=0 pred=0 match=True | acc=0.938 (15/16) | 0.1 samp/s
[17/100] gt=1 pred=0 match=False | acc=0.882 (15/17) | 0.1 samp/s
[18/100] gt=0 pred=0 match=True | acc=0.889 (16/18) | 0.1 samp/s
[19/100] gt=0 pred=0 match=True | acc=0.895 (17/19) | 0.1 samp/s
[20/100] gt=0 pred=0 match=True | acc=0.900 (18/20) | 0.1 samp/s
[21/100] gt=1 pred=1 match=True | acc=0.905 (19/21) | 0.1 samp/s
[22/100] gt=1 pred=1 match=True | acc=0.909 (20/22) | 0.1 samp/s
[23/100] gt=1 pred=0 match=False | acc=0.870 (20/23) | 0.1 samp/s
[24/100] gt=0 pred=0 match=True | acc=0.875 (21/24) | 0.1 samp/s
[25/100] gt=0 pred=0 match=True | acc=0.880 (22/25) | 0.1 samp/s
[26/100] gt=0 pred=0 match=True | acc=0.885 (23/26) | 0.1 samp/s