Commit History

  1. Simplify GeneratorExit handling - let it propagate and rely on finally block

    6dc4bdb
  2. Fix GeneratorExit handling by re-raising after UI reset

    6ff484e
  3. Fix UI state reset after generation cancellation

    fcb2aaa
  4. Fix GeneratorExit exception in cancel generation by adding proper exception handling

    4bc617b
  5. bugfix

    fd70375
  6. Improve cancel generation with robust UI state management and orchestrator pattern

    9ac7f36
  7. Restore cancel generation feature with improved UI integration

    a94befb
  8. Remove cancel generation feature as it didn't work

    efa3ae6
  9. Restore proper cancel functionality⁠

    afa1066
  10. Test stopping criteria effectiveness⁠

    f0d2ab7
  11. Improve cancel generation responsiveness⁠

    48e50b8
  12. Fix cancel generation to gracefully stop ongoing response generation⁠

    a73d8f4
  13. remove qwen3 30b-a3b and qwen3 next 80b-a3b

    4af617b
  14. Remove AOT indication from UI duration estimate

    f97cbfc
  15. Remove AOT compilation completely and enable use_cache

    426163f
  16. Fix dynamic_shapes kwargs to match inputs structure for AOT compilation

    273acf8
  17. Fix AOT compilation dynamic_shapes to match expected arg names for torch.export.export

    0a99dfc
  18. Update requirements.txt

    Verified signature
    4d2c362
  19. Update requirements.txt

    Verified signature
    48d806a
  20. Add dynamic GPU time estimate indicator to UI⁠

    fc989b4
  21. Improve model size detection: replace ad-hoc string parsing with reliable params_b field in MODELS dict

    ab92e0d
  22. Add qwen 80b-a3b

    8cdf3e1
  23. Set better defaults for free-tier users: Qwen3-1.7B model, 1024 max tokens, search disabled

    2cae073
  24. Adjust duration estimation for H200 performance - reduce conservative estimates

    de766da
  25. Use actual parameter count for AOT decision instead of string matching

    e3e334f
  26. Make AOT compilation conditional for models >= 2B parameters to optimize free tier usage

    4500f92
  27. Add AOT compilation optimization for ZeroGPU acceleration

    a7866ff
  28. add 4 20b+ models after enabling dynamic gpu duration

    Verified signature
    fea2910
  29. Add dynamic duration calculation for ZeroGPU acceleration

    6073cc2

  1. make qwen-4b default

    Verified signature
    d3726c6
  2. disable two models that cannot run or too run too slowly on hf spaces with zerogpu

    3dc7ced
  3. Update app.py

    Verified signature
    f1fa55c
  4. Update app.py

    Verified signature
    f07d6ab
  5. Update app.py

    Verified signature
    0992852
  6. add original apriel 15b

    Verified signature
    2b25033
  7. use apriel 8bit

    Verified signature
    a4681bd
  8. run Apriel on 4bit

    Verified signature
    2cadf8a
  9. Add Apriel-1.5-15b-Thinker

    Verified signature
    3665b54

  1. Update app.py

    Verified signature
    7f654b2
  2. Update app.py

    Verified signature
    15b78c7
  3. Update requirements.txt

    Verified signature
    4addcb2
  4. Update requirements.txt

    Verified signature
    49c0c5a
  5. Update requirements.txt

    Verified signature
    831c9e8

  1. Update app.py

    Verified signature
    7356fa6
  2. add 4 models from qwen3 family

    Verified signature
    048cfc4
  3. add qwen3 32b awq

    Verified signature
    b9efb74
  4. Update app.py

    Verified signature
    5e03586
  5. Update README.md

    Verified signature
    c6e816c
  6. Update requirements.txt

    Verified signature
    ff4ad1d
  7. Update app.py

    Verified signature
    e5a1663