Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 5, 2025

• libfuse3 supported split gguf mounting for in-memory split gguf merging

  matbeedotcom

  committed
  a79ad50

  Copy full SHA for a79ad50

Commits on Mar 4, 2025

• ggml : portability fixes for VS 2017 (#12150)

  Show description for 5bbe6a9
  

  mgroeber9110

  and

  Marcus Groeber

  authored
  5bbe6a9

  Copy full SHA for 5bbe6a9

• readme : fix roadmap link (#12185)

  ggerganov

  authored
  20a9b8f

  Copy full SHA for 20a9b8f

• main: allow preloading conversation with -p and add -st / --single-turn (#12145)

  Show description for 56d7a9f
  CISC

  authored
  56d7a9f

  Copy full SHA for 56d7a9f

• `server`: fix deadly typo in response_format.json_schema.schema handling (#12168)

  ochafik

  authored
  1a24c46

  Copy full SHA for 1a24c46

Commits on Mar 3, 2025

• HIP: implement FlashAttention via rocWMMA for CDNA and RDNA3+ (#12032)

  Show description for becade5
  

  3 peopleauthored
  becade5

  Copy full SHA for becade5

• sync : ggml

  Show description for dfd6b2c
  ggerganov

  committed
  dfd6b2c

  Copy full SHA for dfd6b2c

• cuda: unary ops as float + de-duplicate (ggml/1130)

  

  cmdr2

  authored and

  ggerganov

  committed
  b64d7cc

  Copy full SHA for b64d7cc

• sync : ggml

  Show description for 3d1cf3c
  ggerganov

  committed
  3d1cf3c

  Copy full SHA for 3d1cf3c

• cuda/vulkan: specify fp32-only support for some operations in supports_op (ggml/1129)

  Show description for 0cbee13
  

  cmdr2

  authored and

  ggerganov

  committed
  0cbee13

  Copy full SHA for 0cbee13

• sync : ggml

  Show description for 8371d44
  ggerganov

  committed
  8371d44

  Copy full SHA for 8371d44

• cuda/cpu: Increase support for fp16 unary operations (ggml/1125)

  Show description for 87abb7e
  

  cmdr2

  authored and

  ggerganov

  committed
  87abb7e

  Copy full SHA for 87abb7e

• whisper : support GGML_BACKEND_DL (whisper/2843)

  Show description for 6d4c23b
  

  slaren

  and

  ggerganov

  committed
  6d4c23b

  Copy full SHA for 6d4c23b

• cmake : fix compile assumptions for power9/etc (whisper/2777)

  Show description for 6512a90
  

  2 peopleauthored and

  ggerganov

  committed
  6512a90

  Copy full SHA for 6512a90

• Told cmake to install ggml-cpp.h as a public header file. (ggml/1126)

  Show description for 4512055
  

  2 peopleauthored and

  ggerganov

  committed
  4512055

  Copy full SHA for 4512055

• Support pure float16 add/sub/mul/div operations in the CUDA (and CPU) backend (ggml/1121)

  Show description for f54a4ba
  

  cmdr2

  authored and

  ggerganov

  committed
  f54a4ba

  Copy full SHA for f54a4ba

• scripts : sync-ggml-am.sh fix

  ggerganov

  committed
  aede207

  Copy full SHA for aede207

• ci : set GITHUB_ACTION env var for server tests (#12162)

  Show description for 2679c3b
  danbev

  authored
  2679c3b

  Copy full SHA for 2679c3b

• tts: add speaker file support (#12048)

  Show description for c43af92
  

  dm4

  and

  ggerganov

  authored
  c43af92

  Copy full SHA for c43af92

• test-backend-ops : add option -p to filter by op params (#12155)

  slaren

  authored
  d5c63cd

  Copy full SHA for d5c63cd

• ggml : fix kleidiai build (#12159)

  Show description for 9660ffe
  ag2s20150909

  authored
  9660ffe

  Copy full SHA for 9660ffe

• Adding UTF-8 support to llama.cpp (#12111)

  Show description for c950a1f
  ericcurtin

  authored
  c950a1f

  Copy full SHA for c950a1f

• webui : add ?m=... and ?q=... params (#12148)

  Show description for 7b69003
  ngxson

  authored
  7b69003

  Copy full SHA for 7b69003

• SYCL: Move CPY kernels to a separate file and add few missing kernels (#12133)

  Show description for ece9745
  qnixsynapse

  authored
  ece9745

  Copy full SHA for ece9745

Commits on Mar 2, 2025

• ggml-backend : keep paths in native string type when possible (#12144)

  slaren

  authored
  cc473ca

  Copy full SHA for cc473ca

• main: use jinja chat template system prompt by default (#12118)

  Show description for 14dec0c
  

  CISC

  and

  ngxson

  authored
  14dec0c

  Copy full SHA for 14dec0c

Commits on Mar 1, 2025

• main: update outdated system prompt message (followup to #12131) (#12132)

  Show description for 1782cdf
  

  CISC

  and

  ngxson

  authored
  1782cdf

  Copy full SHA for 1782cdf

• common : add --system-prompt parameter, replace behavior of -p in conversation mode (#12131)

  Show description for 45a8e76
  

  CISC

  and

  ngxson

  authored
  45a8e76

  Copy full SHA for 45a8e76

• CUDA: compress mode option and default to size (#12029)

  Show description for 80c41dd
  Green-Sky

  authored
  80c41dd

  Copy full SHA for 80c41dd

• webui : minor typo fixes (#12116)

  Show description for 2cc4a5e
  

  vynride

  and

  ngxson

  authored
  2cc4a5e

  Copy full SHA for 2cc4a5e

Commits on Feb 28, 2025

• convert : fix Norway problem when parsing YAML (#12114)

  Show description for 06c2b15
  ngxson

  authored
  06c2b15

  Copy full SHA for 06c2b15

• ggml : upgrade init_tensor API to return a ggml_status (#11854)

  Show description for 70680c4
  

  WilliamTambellini

  and

  slaren

  authored
  70680c4

  Copy full SHA for 70680c4

• llama : add Phi-4-mini support (supersede #12099) (#12108)

  Show description for c43a3e7
  

  ngxson

  and

  nisparks

  authored
  c43a3e7

  Copy full SHA for c43a3e7

• Update granite vision docs for 3.2 model (#12105)

  Show description for 84d5f4b
  alex-jw-brooks

  authored
  84d5f4b

  Copy full SHA for 84d5f4b

• vulkan: add specific MMV kernels for IQ2 and IQ3 quants + optimizations (#11595)

  Show description for 438a839
  remyoudompheng

  authored
  438a839

  Copy full SHA for 438a839

Pagination

Previous

Next