Commit History

  1. Fix CUDA device-side assert by adjusting max_prompt_length and disabling use_cache

    mayank1365
    fe123ff
  2. Fix IndentationError and duplicate reward logic

    mayank1365
    0d40379
  3. Fix dtype mismatch in training and update blog

    mayank1365
    463c260
  4. fix: resolve tensor size mismatch in reward_function by indexing correctly

    mayank1365
    134ff83
  5. fix: resolve Gradio JSON error by passing dicts directly instead of serialized strings

    mayank1365
    731ed55
  6. fix: resolve 422 errors by sanitizing JSON and fix python syntax error

    mayank1365
    117c380
  7. fix: explicitly add unsloth_zoo to requirements to resolve ImportError

    mayank1365
    f8ddd90
  8. fix: unpin torch and use simpler unsloth install to resolve HF build cache miss

    mayank1365
    091e56d
  9. fix: downgrade CUDA to 12.1 and pin torch versions for HF Space compatibility

    mayank1365
    723a53e
  10. fix: improve GRPO learning signal and handle 422 environment errors

    mayank1365
    f5df9dc
  11. feat: randomize number of facts (2-4) per episode for better training diversity

    mayank1365
    b676c5d
  12. fix: address reward plateau by adding format rewards and improving GRPO logic

    mayank1365
    b588360
  13. fix: resolve GPU detection issue on HF Spaces

    mayank1365
    ef2c4bf

  1. Update app.py

    Verified signature
    69d4d30
  2. Update requirements.txt

    Verified signature
    b8fb5aa
  3. Update requirements.txt

    Verified signature
    392faa7
  4. Update requirements.txt

    Verified signature
    b4a6032
  5. Update requirements.txt

    Verified signature
    47363ae
  6. Create requirements.txt

    Verified signature
    dad4e20
  7. Rename ap.py to app.py

    Verified signature
    01057a2
  8. Rename train_suspect_x.py to ap.py

    Verified signature
    021896a
  9. Upload train_suspect_x.py with huggingface_hub

    Verified signature
    0355697
  10. initial commit

    Verified signature
    1d47818