"vllm/model_executor/models/minicpm.py" did not exist on "21d93c140d0a97af5f0c59e660cf04bd417fd424"
  1. 04 May, 2024 1 commit
    • Michael Goin's avatar
      [Kernel] Support MoE Fp8 Checkpoints for Mixtral (Static Weights with... · 2a052011
      Michael Goin authored
      [Kernel] Support MoE Fp8 Checkpoints for Mixtral (Static Weights with Dynamic/Static Activations) (#4527)
      
      Follow on to #4332 to enable FP8 checkpoint loading for Mixtral and supersedes #4436.
      
      This PR enables the following checkpoint loading features for Mixtral:
      
      Supports loading fp8 checkpoints for Mixtral, such as this "nm-testing/Mixtral-8x7B-Instruct-v0.1-FP8" test model
      Supports static or dynamic activation quantization with static weight quantization (all per tensor)
      Supports different scales for each expert weight
      Supports Fp8 in QKV layer
      Notes:
      
      The Expert Gate/Router always runs at half / full precision for now.
      If there are different weight scales between QKV layer (for separate QKV weights), they are re-quantized using layer.weight_scale.max() so we can have a single gemm for performance.
      2a052011
  2. 03 May, 2024 2 commits
  3. 02 May, 2024 1 commit
  4. 18 Apr, 2024 1 commit
  5. 11 Apr, 2024 2 commits
  6. 03 Apr, 2024 2 commits
  7. 30 Mar, 2024 1 commit
  8. 27 Mar, 2024 1 commit
  9. 25 Mar, 2024 2 commits
  10. 24 Mar, 2024 2 commits
  11. 20 Mar, 2024 1 commit
  12. 16 Mar, 2024 2 commits
  13. 13 Mar, 2024 2 commits
  14. 11 Mar, 2024 1 commit
  15. 07 Mar, 2024 1 commit
  16. 27 Feb, 2024 1 commit
  17. 22 Feb, 2024 1 commit
  18. 06 Feb, 2024 2 commits
  19. 05 Feb, 2024 1 commit
  20. 01 Feb, 2024 1 commit
  21. 31 Jan, 2024 2 commits
  22. 30 Jan, 2024 2 commits
  23. 29 Jan, 2024 1 commit
  24. 22 Jan, 2024 1 commit
  25. 18 Jan, 2024 1 commit
  26. 14 Jan, 2024 1 commit
  27. 04 Jan, 2024 1 commit
  28. 03 Jan, 2024 2 commits
  29. 10 Dec, 2023 1 commit