1. 09 Dec, 2025 1 commit
  2. 29 Nov, 2025 1 commit
  3. 20 Nov, 2025 1 commit
  4. 13 Nov, 2025 1 commit
  5. 10 Nov, 2025 1 commit
  6. 30 Oct, 2025 1 commit
  7. 23 Oct, 2025 1 commit
  8. 12 Oct, 2025 1 commit
  9. 05 Oct, 2025 1 commit
  10. 03 Oct, 2025 1 commit
  11. 26 Sep, 2025 1 commit
  12. 19 Sep, 2025 1 commit
  13. 17 Sep, 2025 1 commit
  14. 15 Sep, 2025 1 commit
  15. 12 Sep, 2025 1 commit
  16. 14 Aug, 2025 1 commit
  17. 23 Jul, 2025 1 commit
  18. 19 Jun, 2025 1 commit
  19. 03 Jun, 2025 1 commit
  20. 23 May, 2025 1 commit
  21. 12 May, 2025 1 commit
  22. 04 May, 2025 1 commit
  23. 26 Apr, 2025 1 commit
  24. 24 Apr, 2025 1 commit
  25. 22 Apr, 2025 1 commit
  26. 17 Apr, 2025 1 commit
  27. 29 Mar, 2025 1 commit
  28. 24 Mar, 2025 2 commits
  29. 13 Mar, 2025 1 commit
  30. 12 Mar, 2025 1 commit
  31. 10 Mar, 2025 1 commit
  32. 06 Mar, 2025 1 commit
  33. 03 Mar, 2025 3 commits
  34. 25 Feb, 2025 1 commit
  35. 12 Feb, 2025 1 commit
  36. 11 Feb, 2025 1 commit
  37. 07 Feb, 2025 1 commit
    • afeldman-nm's avatar
      [V1] Logprobs and prompt logprobs support (#9880) · 0630d453
      afeldman-nm authored
      
      
      This PR is adding support for sample logprobs & prompt logprobs to vLLM v1.
      
      New behavior:
      
      - During model execution, model runner computes sample logprobs (if user-provided logprobs setting is not None) and prompt logprobs (if user-provided prompt_logprobs setting is not None). For both sample and prompt logprobs, the engine core returns 3 vectors: token ids, token logprob values, token ranks. Ranks reflect tokens' 1-indexed positions in the vocabulary vector after sorting the vocabulary by log probability in descending order.
      - In scheduler.update_from_output(), sample and prompt logprobs are incorporated into the EngineCoreOutput data structure which is transferred to the engine client. If multiprocessing is enabled, then sample and prompt logprobs will be (de)serialized when the EngineCoreOutput data structure is (de)serialized.
      - During output processing, the LogprobsProcessor transforms the triplet of token ids, token logprobs values, and token ranks into the OpenAI-compatible List[Dict[token id,Logprob]] format (for sample and prompt logprobs respectively.)
      - Each Logprob instance (whether sample- or prompt-) consists of a token's log-probability, rank, and detokenized string representation. Note that logprob detokenization is handled by the LogprobsProcessor not the detokenizer.
      Signed-off-by: default avatarAndrew Feldman <afeldman@neuralmagic.com>
      Signed-off-by: default avatarNick Hill <nhill@redhat.com>
      Signed-off-by: default avatarrshaw@neuralmagic.com <rshaw@neuralmagic.com>
      Co-authored-by: default avatarrshaw@neuralmagic.com <rshaw@neuralmagic.com>
      Co-authored-by: default avatarNick Hill <nhill@redhat.com>
      0630d453