Commit History

  1. Upload folder using huggingface_hub

    Verified signature
    1e65f1e
  2. Upload folder using huggingface_hub

    Verified signature
    fa2d96c

  1. Upload folder using huggingface_hub

    Verified signature
    17163ae

  1. Update Blog.md

    Verified signature
    629b1ef
  2. Upload folder using huggingface_hub

    Verified signature
    322a982
  3. Upload folder using huggingface_hub

    Verified signature
    3edbdb5
  4. Upload folder using huggingface_hub

    Verified signature
    b30ff25
  5. Upload folder using huggingface_hub

    Verified signature
    9b34b37
  6. grpo: add --eval_best_of (best-of-N eval for SFT and GRPO)

    Verified signature
    8a2235b
  7. grpo: add --eval_best_of (best-of-N eval for SFT and GRPO)

    Verified signature
    97a5555
  8. GRPO v3_anchored: fix promote mode-collapse

    Verified signature
    e10110e
  9. GRPO v2_delta_phase reward + diversified states + tuned smoke schedule

    Verified signature
    da8eeea
  10. GRPO v2_delta_phase reward + diversified states + tuned smoke schedule

    Verified signature
    715e7e0
  11. Colab: HF Jobs orchestrator (full 400/200/3 run)

    Verified signature
    9dcee98
  12. Update usage comment for full training

    Verified signature
    04a98b0
  13. Bump default max_steps to 200

    Verified signature
    fe2d531
  14. Fix notebook: unsloth-first imports, HF spaces clone, GPU check

    Verified signature
    86e410d
  15. Add Colab notebook for full GRPO training

    Verified signature
    009f50c
  16. Patch missing warnings_issued attr for transformers 5.x compat

    Verified signature
    9eb8e28
  17. Tighten pydantic pin to <2.11 to match mergekit ~=2.10.6

    Verified signature
    1994102
  18. Pin pydantic <2.12 after editable install to fix mergekit compat

    Verified signature
    80f9fa1
  19. Add weave dep (another TRL 0.24 hard import in callbacks.py)

    Verified signature
    fdd1260
  20. Patch TRANSFORMERS_CACHE in sanity check too

    Verified signature
    d29d073
  21. Patch TRANSFORMERS_CACHE for llm_blender compat

    Verified signature
    83ee735
  22. Add llm-blender dep required by trl 0.24 judges

    Verified signature
    fad96c9
  23. Add mergekit dep required by trl 0.24 callbacks

    Verified signature
    6ad440d
  24. Fix total_mem -> total_memory in GPU check

    Verified signature
    d24665d
  25. Add GRPO shell launcher with pinned deps

    Verified signature
    fcdfb56
  26. Add standalone GRPO training script

    Verified signature
    81863b2
  27. Fix GRPO smoke dependencies with llm-blender

    Verified signature
    f63b625
  28. Fix GRPO smoke dependencies with llm-blender

    Verified signature
    8ef8d11
  29. Fix GRPO training script import path

    Verified signature
    a8c1549
  30. Add PYTHONPATH check to GRPO smoke launcher

    Verified signature
    2bd1cf1
  31. Fix GRPO training script import path

    Verified signature
    04307dd
  32. Fix GRPO smoke dependencies with llm-blender

    Verified signature
    bb75a43
  33. Pin TRL version for GRPO smoke job

    Verified signature
    b675479
  34. Fix GRPO smoke job by installing mergekit

    Verified signature
    9b04641
  35. Fix GRPO smoke job torchao incompatibility

    Verified signature
    d4fe404
  36. Add GRPO smoke job launcher

    Verified signature
    9c5dd45
  37. Add GRPO smoke job launcher

    Verified signature
    a4c81e5

  1. Add GRPO training script for DesignGym 2.0

    Verified signature
    df2f172
  2. Upload folder using huggingface_hub

    Verified signature
    05acbb9
  3. Upload folder using huggingface_hub

    Verified signature
    ec45357
  4. Upload folder using huggingface_hub

    Verified signature
    7b90e91
  5. Upload folder using huggingface_hub

    Verified signature
    44c2d9e

  1. Upload folder using huggingface_hub

    Verified signature
    f48180b
  2. Upload folder using huggingface_hub

    Verified signature
    1b32a8c
  3. Upload folder using huggingface_hub

    Verified signature
    f95e1d0
  4. Upload folder using huggingface_hub

    Verified signature
    566fbb4
  5. Upload folder using huggingface_hub

    Verified signature
    73f77c2