Commit History

Автор SHA1 Съобщение Дата
  royjhan b9f5e16c80 Introduce `/api/embed` endpoint supporting batch embedding (#5127) преди 9 месеца
  Jeffrey Morgan ef98803d63 llm: looser checks for minimum memory (#5677) преди 9 месеца
  Jeffrey Morgan c4cf8ad559 llm: avoid loading model if system memory is too small (#5637) преди 9 месеца
  Jeffrey Morgan 791650ddef sched: only error when over-allocating system memory (#5626) преди 9 месеца
  Daniel Hiltgen 22c81f62ec Remove duplicate merge glitch преди 9 месеца
  Michael Yang 9bbddc37a7 Merge pull request #5126 from ollama/mxyng/messages преди 9 месеца
  Jeffrey Morgan 53da2c6965 llm: remove ambiguous comment when putting upper limit on predictions to avoid infinite generation (#5535) преди 9 месеца
  Michael Yang ac7a842e55 fix model reloading преди 10 месеца
  Daniel Hiltgen ccd7785859 Merge pull request #5243 from dhiltgen/modelfile_use_mmap преди 10 месеца
  Daniel Hiltgen 0e982bc1f4 Fix corner cases on tmp cleaner on mac преди 10 месеца
  Josh Yan 33a65e3ba3 error преди 10 месеца
  Daniel Hiltgen 97c9e11768 Switch use_mmap to a pointer type преди 10 месеца
  Daniel Hiltgen 3518aaef33 Merge pull request #4218 from dhiltgen/auto_parallel преди 10 месеца
  Blake Mizerany cb42e607c5 llm: speed up gguf decoding by a lot (#5246) преди 10 месеца
  Daniel Hiltgen 17b7186cd7 Enable concurrency by default преди 1 година
  Daniel Hiltgen 5bf5aeec01 Refine mmap default logic on linux преди 10 месеца
  Daniel Hiltgen 96624aa412 Merge pull request #5072 from dhiltgen/windows_path преди 10 месеца
  Daniel Hiltgen 7784ca33ce Tighten up memory prediction logging преди 10 месеца
  Daniel Hiltgen 171796791f Adjust mmap logic for cuda windows for faster model load преди 10 месеца
  Daniel Hiltgen b2799f111b Move libraries out of users path преди 10 месеца
  Daniel Hiltgen da3bf23354 Workaround gfx900 SDMA bugs преди 11 месеца
  Daniel Hiltgen 6f351bf586 review comments and coverage преди 11 месеца
  Daniel Hiltgen fc37c192ae Refine CPU load behavior with system memory visibility преди 11 месеца
  Daniel Hiltgen 6fd04ca922 Improve multi-gpu handling at the limit преди 11 месеца
  Craig Hughes b84aea1685 Critical fix from llama.cpp JSON grammar to forbid un-escaped escape characters inside strings, which breaks parsing. (#3782) преди 10 месеца
  Michael Yang e40145a39d lint преди 11 месеца
  Michael Yang c895a7d13f some gocritic преди 11 месеца
  Michael Yang 829ff87bd1 revert tokenize ffi (#4761) преди 11 месеца
  Jeffrey Morgan a50a87a7b8 partial offloading: allow flash attention and disable mmap (#4734) преди 11 месеца
  Michael Yang 26a00a0410 use ffi for tokenizing/detokenizing преди 11 месеца