Commits

Branch selector

 a410a9e

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 13, 2023

• unused change reverted

  FSSRepo

  committed
  a410a9e

  Copy full SHA for a410a9e

• server ui now support multiple clients

  FSSRepo

  committed
  6358ae5

  Copy full SHA for 6358ae5

• chat.mjs support cached prompt + some fixes

  FSSRepo

  committed
  4ba5a50

  Copy full SHA for 4ba5a50

• cached prompt support

  FSSRepo

  committed
  500ac71

  Copy full SHA for 500ac71

Commits on Oct 12, 2023

• grammar + no stream completion

  FSSRepo

  committed
  83c2b35

  Copy full SHA for 83c2b35

• multiple client support

  FSSRepo

  committed
  5b8e29d

  Copy full SHA for 5b8e29d

• completion endpoint working

  FSSRepo

  committed
  8148480

  Copy full SHA for 8148480

• refactored sampling function

  FSSRepo

  committed
  29c8cdd

  Copy full SHA for 29c8cdd

• Merge branch 'master' of https://github.com/ggerganov/llama.cpp

  FSSRepo

  committed
  b716eeb

  Copy full SHA for b716eeb

• save dev progress

  FSSRepo

  committed
  7850421

  Copy full SHA for 7850421

• examples: support LLaVA v1.5 (multimodal model) (#3436)

  Show description for 370359e
  

  monatis

  and

  ggerganov

  authored
  370359e

  Copy full SHA for 370359e

• docs : fix typo GOMP_CPU_AFFINITY (#3597)

  maekawatoshiki

  authored
  9e24cc6

  Copy full SHA for 9e24cc6

• cmake : fix add_compile_options on macOS

  ggerganov

  committed
  d28e572

  Copy full SHA for d28e572

• typo : it is --n-gpu-layers not --gpu-layers (#3592)

  Show description for f3040be
  ianscrivener

  authored
  f3040be

  Copy full SHA for f3040be

• ci : check if there is enough VRAM (#3596)

  Show description for 1a8c879
  ggerganov

  authored
  1a8c879

  Copy full SHA for 1a8c879

• server : add completion mode (no chat) (#3582)

  akx

  authored
  b016596

  Copy full SHA for b016596

• prompts : add mnemonics.txt

  ggerganov

  committed
  6b3ae4d

  Copy full SHA for 6b3ae4d

• server : fix kv cache management (#3588)

  ggerganov

  authored
  57dd55e

  Copy full SHA for 57dd55e

Commits on Oct 11, 2023

• crash fixed

  FSSRepo

  committed
  4712302

  Copy full SHA for 4712302

• implementing parallel decoding in server example

  FSSRepo

  committed
  63f99b1

  Copy full SHA for 63f99b1

• main : fix session loading bug (#3400)

  ggerganov

  committed
  b8fe4b5

  Copy full SHA for b8fe4b5

• server : add parameter -tb N, --threads-batch N (#3584)

  Show description for a8bdd65
  

  m18coppola

  and

  Michael Coppola

  authored
  a8bdd65

  Copy full SHA for a8bdd65

• common : fix mirostat state when using multiple sequences (#3543)

  Show description for 70c29da
  KerfuffleV2

  authored
  70c29da

  Copy full SHA for 70c29da

• batched : add bench tool (#3545)

  Show description for 8c70a5f
  ggerganov

  authored
  8c70a5f

  Copy full SHA for 8c70a5f

• examples : add batched.swift + improve CI for swift (#3562)

  zshannon

  authored
  24ba3d8

  Copy full SHA for 24ba3d8

Commits on Oct 10, 2023

• Add MPT model to supported models in README.md (#3574)

  Galunid

  authored
  9f6ede1

  Copy full SHA for 9f6ede1

• Minor improvements in GPT2 tokenizer (#3567)

  Show description for 233fc1c
  goerch

  authored
  233fc1c

  Copy full SHA for 233fc1c

• readme : add bloom (#3570)

  xingchensong

  authored
  c5b4936

  Copy full SHA for c5b4936

• llm : add bloom models (#3553)

  Show description for 02d2875
  

  xingchensong

  and

  ggerganov

  authored
  02d2875

  Copy full SHA for 02d2875

• swift : improvements and fixes (#3564)

  Show description for 0aa6595
  jhen0409

  authored
  0aa6595

  Copy full SHA for 0aa6595

• llm : add MPT support (#3417)

  Show description for f5f9121
  

  3 peopleauthored
  f5f9121

  Copy full SHA for f5f9121

• infill. : fix tokenization (#3508)

  Show description for 11ea5c7
  vvhg1

  authored
  11ea5c7

  Copy full SHA for 11ea5c7

Commits on Oct 9, 2023

• ggml-alloc : fix assert in debug builds (#3555)

  slaren

  authored
  95bd60a

  Copy full SHA for 95bd60a

• refact : fix convert script + zero out KV cache to avoid nans (#3523)

  Show description for fcca0a7
  ggerganov

  authored
  fcca0a7

  Copy full SHA for fcca0a7

• metal : do not use mul_mm kernels when ne00 < 64 (#3542)

  ggerganov

  authored
  dcc09d2

  Copy full SHA for dcc09d2

Pagination

Previous

Next