##### f28: SFT from base on ex/full.jsonl
{'train_runtime': 659.9466, 'train_samples_per_second': 86.807, 'train_steps_per_second': 2.715, 'train_loss': 0.02382280319995646, 'epoch': 4.0}
saved -> f28
The tokenizer you are loading from '/root/mt92/f28' 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.
The following generation flags are not valid and may be ignored: ['temperature', 'top_p', 'top_k']. Set `TRANSFORMERS_VERBOSITY=info` for more details.
=== f28 on val_all.jsonl (4710 docs, 192s) ===
  MICRO-F1 0.9942 | P 0.9966 R 0.9918 | tp/fp/fn 8845/30/73 | malformed 3 | out tok med 34 max 256
The tokenizer you are loading from '/root/mt92/f28' 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.
The following generation flags are not valid and may be ignored: ['temperature', 'top_p', 'top_k']. Set `TRANSFORMERS_VERBOSITY=info` for more details.
=== f28 on test_all.jsonl (5017 docs, 205s) ===
  MICRO-F1 0.9944 | P 0.9959 R 0.9928 | tp/fp/fn 9089/37/66 | malformed 2 | out tok med 33 max 256
The tokenizer you are loading from '/root/mt92/f28' 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.
The following generation flags are not valid and may be ignored: ['temperature', 'top_p', 'top_k']. Set `TRANSFORMERS_VERBOSITY=info` for more details.
=== f28 on gen_holdout.jsonl (300 docs, 24s) ===
  MICRO-F1 0.9404 | P 0.9396 R 0.9413 | tp/fp/fn 529/34/33 | malformed 0 | out tok med 34 max 192
STAGE1_DONE