Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Jan 4, 2025

• Vulkan: Add device-specific blacklist for coopmat for the AMD proprietary driver (#11074)

  Show description for b56f079
  0cc4m

  authored
  b56f079

• llama : Add support for DeepSeek V3 (#11049)

  Show description for 9394bbd
  

  fairydreaming

  and

  sszymczy

  authored
  9394bbd

• [GGML][RPC] Support for models with non-512-aligned tensors over RPC. (#11047)

  Show description for f922a9c
  

  matt23654

  and

  slaren

  authored
  f922a9c

• llama : add support for the cohere2 model architecture (#10900)

  dranger003

  authored
  46be942

• sync : ggml

  ggerganov

  committed
  78c6785

• ggml : do not install metal source when embed library (ggml/1054)

  ggerganov

  committed
  5e3b08d

• ggml : improve inputs log sched_print_assignments (ggml/1053)

  Show description for db68c93
  

  danbev

  authored and

  ggerganov

  committed
  db68c93

• fix: Vulkan shader gen binary path (#11037)

  giladgd

  authored
  c31fc8b

Commits on Jan 3, 2025

• common : disable KV cache shifting automatically for unsupported models (#11053)

  Show description for 4b0c638
  

  MollySophia

  and

  ggerganov

  authored
  4b0c638

• metal : avoid uint (#11019)

  ggerganov

  committed
  e7da954

• llama : refactor src/llama.cpp (#10902)

  Show description for f66f582
  ggerganov

  authored
  f66f582

Commits on Jan 2, 2025

• server: bench: minor fixes (#10765)

  Show description for 2f0ee84
  phymbert

  authored
  2f0ee84

• server : allow using LoRA adapters per-request (#10994)

  Show description for 0da5d86
  

  ngxson

  and

  ggerganov

  authored
  0da5d86

• readme : add llama-swap to infrastructure section (#11032)

  Show description for a45433b
  mostlygeek

  authored
  a45433b

Commits on Dec 31, 2024

• ggml : fixes for AVXVNNI instruction set with MSVC and Clang (#11027)

  Show description for 0827b2c
  

  Srihari-mcw

  and

  slaren

  authored
  0827b2c

• server : clean up built-in template detection (#11026)

  Show description for 45095a6
  ngxson

  authored
  45095a6

• server : add OAI compat for /v1/completions (#10974)

  Show description for 5896c65
  ngxson

  authored
  5896c65

• convert : fix Llama-3_1-Nemotron-51B rope settings (#11008)

  Show description for bc7b1f8
  ymcki

  authored
  bc7b1f8

• common, examples, ggml : fix MSYS2 GCC compiler errors and warnings when building with LLAMA_CURL=ON and GGML_OPENCL=ON (#11013)

  Show description for 6e1531a
  peter277

  authored
  6e1531a

Commits on Dec 30, 2024

• vulkan: optimize mul_mat for small values of N (#10991)

  Show description for 716bd6d
  jeffbolznv

  authored
  716bd6d

• android : fix llama_batch free (#11014)

  ag2s20150909

  authored
  c250ecb

Commits on Dec 29, 2024

• vulkan: im2col and matmul optimizations for stable diffusion (#10942)

  Show description for a813bad
  jeffbolznv

  authored
  a813bad

• vulkan: Use push constant offset to handle misaligned descriptors (#10987)

  jeffbolznv

  authored
  fdd2188

Commits on Dec 28, 2024

• server: added more docs for response_fields field (#10995)

  isaac-mcfadyen

  authored
  f865ea1

• server : fix token duplication when streaming with stop strings (#10997)

  z80maniac

  authored
  16cdce7

Commits on Dec 26, 2024

• vulkan: multi-row k quants (#10846)

  Show description for d79d8f3
  netrunnereve

  authored
  d79d8f3

• examples, ggml : fix GCC compiler warnings (#10983)

  Show description for d283d02
  peter277

  authored
  d283d02

Commits on Dec 24, 2024

• server : add support for "encoding_format": "base64" to the */embeddings endpoints (#10967)

  Show description for 9ba399d
  

  elk-cloner

  and

  ngxson

  authored
  9ba399d

• ggml : more perfo with llamafile tinyblas on x86_64 (#10714)

  Show description for 2cd43f4
  Djip007

  authored
  2cd43f4

• server: allow filtering llama server response fields (#10940)

  Show description for 09fe2e7
  

  nvrxq

  and

  ngxson

  authored
  09fe2e7

• llama : the WPM vocabs use the CLS token as BOS (#10930)

  Show description for 30caac3
  ggerganov

  authored
  30caac3

• ggml : use wstring for backend search paths (#10960)

  Show description for 60cfa72
  slaren

  authored
  60cfa72

• ggml : fix arm enabled features check (#10961)

  slaren

  authored
  3327bb0

Commits on Dec 23, 2024

• ggml : fix const usage in SSE path (#10962)

  slaren

  authored
  32d6ee6

• server : fix missing model id in /model endpoint (#10957)

  Show description for 14b699e
  ngxson

  authored
  14b699e

Pagination

PreviousNext