Cronologia Commit

Autore SHA1 Messaggio Data
  AlpinDale 60ca1e1e5e feat: add ngram prompt lookup decoding for speculative decoding (#438) 9 mesi fa
  AlpinDale d8c4193704 feat: Speculative Decoding using a draft model (#432) 9 mesi fa
  AlpinDale 76f36af704 feat: LM Format Enforcer support (#428) 9 mesi fa
  AlpinDale bd0ddf1cfe feat: EETQ quantization (#408) 10 mesi fa
  AlpinDale 4d33ce60da feat: Triton flash attention backend for ROCm (#407) 10 mesi fa
  AlpinDale 071269e406 feat: FP8 E4M3 KV Cache (#405) 10 mesi fa
  AlpinDale 6f00203041 refactor scheduler for chunked prefill, remove reorder policy for now 10 mesi fa
  AlpinDale 9aaeb5d349 add speculative config and arg for later 10 mesi fa
  AlpinDale a304f76d89 feat: Intel CPU support (#403) 10 mesi fa
  AlpinDale fa083286e3 Speculative Decoding Part 4: Lookahead scheduling (#402) 10 mesi fa
  AlpinDale f845a661dd Chunked Prefill Part 2: data update 10 mesi fa
  AlpinDale eff5eb16c5 ruff 10 mesi fa
  AlpinDale 753f6dc51b add v2 block manager 10 mesi fa
  AlpinDale 7b9c08afae vision model support 10 mesi fa
  AlpinDale b738554558 add reorder scheduler policy 10 mesi fa
  AlpinDale 1ba9ff78cd add scheduler delay factor 10 mesi fa
  AlpinDale 78d66f16d1 Chunked Prefill Part 1 (#384) 10 mesi fa
  AlpinDale feb5840f2a feat: async tokenization (#374) 10 mesi fa
  AlpinDale f8dfac6372 chore: attention refactor and upstream sync apr01 (#365) 10 mesi fa
  AlpinDale c41462cfcd feat: exllamav2 quantization (#305) 11 mesi fa
  AlpinDale 9810daa699 feat: INT8 KV Cache (#298) 11 mesi fa
  AlpinDale e0c35bb353 feat: bitsandbytes and `--load-in{4,8}bit` support (#294) 11 mesi fa
  AlpinDale 705821a7fe feat: AQLM quantization support (#293) 11 mesi fa
  AlpinDale ac82b67f75 feat: naive context shift and various QoL changes (#289) 11 mesi fa
  AlpinDale 72229a94da feat: better marlin kernels (#285) 11 mesi fa
  AlpinDale 657aec0cbd refactor: OpenAI endpoint (#261) 11 mesi fa
  AlpinDale 4d04ade9ef feat: fine-grained seeds (#279) 11 mesi fa
  AlpinDale ea0f57b233 feat: allow further support for non-cuda devices (#247) 1 anno fa
  AlpinDale c3a221eb02 feat: GGUF, QuIP#, and Marlin support (#228) 1 anno fa
  AlpinDale 31c95011a6 feat: FP8 E5M2 KV Cache (#226) 1 anno fa