Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 21, 2025

• ggml : add SSE 4.2 and x64 base variant for CPUs without AVX (#12871)

  Show description for 1d735c0
  slaren

  authored
  1d735c0

  Copy full SHA for 1d735c0

• SYCL: Add non-contiguous support in ROPE (#12993)

  Show description for 5368ddd
  qnixsynapse

  authored
  5368ddd

  Copy full SHA for 5368ddd

• mtmd : merge llava, gemma3 and minicpmv CLI into single `llama-mtmd-cli` (#13012)

  Show description for 84a9bf2
  

  ngxson

  and

  ggerganov

  authored
  84a9bf2

  Copy full SHA for 84a9bf2

Commits on Apr 20, 2025

• convert : experimental support for `--mmproj` flag (#13023)

  Show description for 2016f07
  

  ngxson

  and

  compilade

  authored
  2016f07

  Copy full SHA for 2016f07

• llava: fix errors in clip.h on certain compilers (#13030)

  jmorganca

  authored
  6602304

  Copy full SHA for 6602304

• vulkan: support noncontiguous rms_norm (#13031)

  jeffbolznv

  authored
  6616820

  Copy full SHA for 6616820

• metal: add neg operator (#13029)

  jmorganca

  authored
  4ba9d71

  Copy full SHA for 4ba9d71

Commits on Apr 19, 2025

• Disable CI cross-compile builds (#13022)

  bandoti

  authored
  0013715

  Copy full SHA for 0013715

• gguf-py : fix upload python package workflow (#13020)

  CISC

  authored
  fb28f4f

  Copy full SHA for fb28f4f

• clip : refactor, add `image_manipulation` and `llava_uhd` classes (#13011)

  Show description for 37b9f0d
  ngxson

  authored
  37b9f0d

  Copy full SHA for 37b9f0d

Commits on Apr 18, 2025

• main : Fix Ctrl+D/newline handling (#12951)

  Show description for 6408210
  danielzgtg

  authored
  6408210

  Copy full SHA for 6408210

• gguf-py : GGUF Editor GUI - Python + Qt6 (#12930)

  christopherthompson81

  authored
  aff9d10

  Copy full SHA for aff9d10

• server : use std::move whenever possible (#12936)

  Show description for 35370ba
  

  ngxson

  and

  ggerganov

  authored
  35370ba

  Copy full SHA for 35370ba

• SYCL: Refactor and enable FP16 in binary broadcast OPs (#12975)

  Show description for 8d66005
  qnixsynapse

  authored
  8d66005

  Copy full SHA for 8d66005

• mtmd : add methods to access `mtmd_image_tokens` (#12906)

  Show description for b9154ec
  ngxson

  authored
  b9154ec

  Copy full SHA for b9154ec

• rpc : add RPC_CMD_HELLO (#12955)

  Show description for 2db9ba1
  rgerganov

  authored
  2db9ba1

  Copy full SHA for 2db9ba1

Commits on Apr 17, 2025

• graph : make FA compatible with MLA + add initial Metal kernels (#12953)

  Show description for 2f74c35
  ggerganov

  authored
  2f74c35

  Copy full SHA for 2f74c35

• ggml: Re-enable CUDA graphs in presence of CONT and DUP nodes (#12970)

  agray3

  authored
  207c22e

  Copy full SHA for 207c22e

• CANN: Add support for async operator submission (#12864)

  Show description for 7a395f6
  hipudding

  authored
  7a395f6

  Copy full SHA for 7a395f6

• llama : recognize IBM Granite 3.3 FIM tokens (#12988)

  Show description for 971f245
  Noeda

  authored
  971f245

  Copy full SHA for 971f245

Commits on Apr 16, 2025

• opencl: fix incorrect local_size index in profiling log (#12868)

  kimminsu38oo

  authored
  12b1750

  Copy full SHA for 12b1750

• vulkan: enable coopmat2 FA gqa and split_k optimizations more often (#12931)

  Show description for 015022b
  jeffbolznv

  authored
  015022b

  Copy full SHA for 015022b

• CANN: Add 310P operator support check (#12962)

  noemotiovon

  authored
  b43d89e

  Copy full SHA for b43d89e

Commits on Apr 15, 2025

• opencl: split `ggml-opencl.cl` into multiple files and cleanup (#12886)

  Show description for 80f19b4
  

  lhez

  and

  shawngu-quic

  authored
  80f19b4

  Copy full SHA for 80f19b4

• metal : add FA-vec kernels for head size 96 (#12952)

  Show description for f8f820c
  ggerganov

  authored
  f8f820c

  Copy full SHA for f8f820c

• CANN: Add x86 build ci (#12950)

  Show description for 54a7272
  hipudding

  authored
  54a7272

  Copy full SHA for 54a7272

• CUDA/HIP: Share the same unified memory allocation logic. (#12934)

  Show description for 84778e9
  hjc4869

  authored
  84778e9

  Copy full SHA for 84778e9

• SYCL: Add ROPE vision kernel (#12887)

  Show description for 5106764
  qnixsynapse

  authored
  5106764

  Copy full SHA for 5106764

• llama : DeepSeek V2/V3 MLA implementation (#12801)

  Show description for daa4228
  jukofyork

  authored
  daa4228

  Copy full SHA for daa4228

• ggml : Add AVX512 implementation of GEMM - Q4_Kx8 (#12829)

  Show description for eccc7a1
  Srihari-mcw

  authored
  eccc7a1

  Copy full SHA for eccc7a1

• CANN: Opt ROPE optimization (#12865)

  Show description for 0019279
  noemotiovon

  authored
  0019279

  Copy full SHA for 0019279

• CANN: Optimize CANN buffer pool memory management (#12875)

  Show description for b0c75ac
  bachelor-dou

  authored
  b0c75ac

  Copy full SHA for b0c75ac

Commits on Apr 14, 2025

• Add performance print for gemma3 in example (#12929)

  Russyyds

  authored
  d6d2c2a

  Copy full SHA for d6d2c2a

• SYCL: Fix im2col (#12910)

  Show description for 75afa0a
  qnixsynapse

  authored
  75afa0a

  Copy full SHA for 75afa0a

• rpc : use ggml_context_ptr (#12938)

  rgerganov

  authored
  c772d54

  Copy full SHA for c772d54

Pagination

Previous

Next