Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 5, 2023

• ggml-cuda : fix f16 mul mat (#3961)

  Show description for 2833a6f
  slaren

  authored
  2833a6f

  Copy full SHA for 2833a6f

• Allow common process_escapes to handle \x sequences (#3928)

  Show description for d9ccce2
  KerfuffleV2

  authored
  d9ccce2

  Copy full SHA for d9ccce2

• server : fix typo for --alias shortcut from -m to -a (#3958)

  RoyalHeart

  authored
  bb60fd0

  Copy full SHA for bb60fd0

• cuda : fix disabling device with --tensor-split 1,0 (#3951)

  Show description for 132d25b
  

  cebtenzzre

  and

  slaren

  authored
  132d25b

  Copy full SHA for 132d25b

• llama : mark LLM_ARCH_STARCODER as full offload supported (#3945)

  Show description for 3d48f42
  wsxiaoys

  authored
  3d48f42

  Copy full SHA for 3d48f42

• cmake : MSVC instruction detection (fixed up #809) (#3923)

  Show description for c41ea36
  

  3 peopleauthored
  c41ea36

  Copy full SHA for c41ea36

• ci : use intel sde when ci cpu doesn't support avx512 (#3949)

  netrunnereve

  authored
  a7fac01

  Copy full SHA for a7fac01

• cuda : revert CUDA pool stuff (#3944)

  Show description for 48ade94
  slaren

  authored
  48ade94

  Copy full SHA for 48ade94

Commits on Nov 4, 2023

• gguf-py: Support 01.AI Yi models (#3943)

  KerfuffleV2

  authored
  f28af0d

  Copy full SHA for f28af0d

Commits on Nov 3, 2023

• metal : round up to 16 to fix MTLDebugComputeCommandEncoder assertion (#3938)

  psugihara

  authored
  d9b33fe

  Copy full SHA for d9b33fe

• ggml-metal: fix yarn rope (#3937)

  jxy

  authored
  5ba3746

  Copy full SHA for 5ba3746

• ggml-cuda : move row numbers to x grid dim in mmv kernels (#3921)

  slaren

  authored
  abb77e7

  Copy full SHA for abb77e7

• speculative : change default p_accept to 0.5 + CLI args (#3919)

  Show description for 8f961ab
  ggerganov

  committed
  8f961ab

  Copy full SHA for 8f961ab

• common : YAYF (yet another YARN fix) (#3925)

  Show description for 0581602
  ggerganov

  authored
  0581602

  Copy full SHA for 0581602

• llama : change yarn_ext_factor placeholder to -1 (#3922)

  cebtenzzre

  authored
  3fdbe6b

  Copy full SHA for 3fdbe6b

Commits on Nov 2, 2023

• cuda : add ROCM aliases for CUDA pool stuff (#3918)

  KerfuffleV2

  authored
  629f917

  Copy full SHA for 629f917

• cmake : fix relative path to git submodule index (#3915)

  abetlen

  authored
  51b2fc1

  Copy full SHA for 51b2fc1

• readme : add notice about #3912

  ggerganov

  authored
  224e7d5

  Copy full SHA for 224e7d5

• cuda : fix const ptrs warning causing ROCm build issues (#3913)

  ggerganov

  authored
  c7743fe

  Copy full SHA for c7743fe

• cuda : use CUDA memory pool with async memory allocation/deallocation when available (#3903)

  Show description for d606905
  

  young-developer

  and

  Oleksii Maryshchenko

  authored
  d606905

  Copy full SHA for d606905

• gguf : print error for GGUFv1 files (#3908)

  ggerganov

  authored
  4ff1046

  Copy full SHA for 4ff1046

• cmake : disable LLAMA_NATIVE by default (#3906)

  slaren

  authored
  21958bb

  Copy full SHA for 21958bb

• gguf : remove special-case code for GGUFv1 (#3901)

  Show description for 2756c4f
  ggerganov

  authored
  2756c4f

  Copy full SHA for 2756c4f

• llm : prevent from 1-D tensors being GPU split (#3697)

  ggerganov

  committed
  1efae9b

  Copy full SHA for 1efae9b

• build : link against build info instead of compiling against it (#3879)

  Show description for b12fa0d
  

  cebtenzzre

  and

  kassane

  authored
  b12fa0d

  Copy full SHA for b12fa0d

• cuda : check if this fixes Pascal card regression (#3882)

  ggerganov

  authored
  4d719a6

  Copy full SHA for 4d719a6

• metal : fix build errors and kernel sig after #2268 (#3898)

  ggerganov

  authored
  183b3fa

  Copy full SHA for 183b3fa

• cuda : fix RoPE after #2268 (#3897)

  cebtenzzre

  authored
  2fffa0d

  Copy full SHA for 2fffa0d

Commits on Nov 1, 2023

• llama : fix llama_context_default_params after #2268 (#3893)

  cebtenzzre

  authored
  0eb332a

  Copy full SHA for 0eb332a

• ggml-cuda : compute ptrs for cublasGemmBatchedEx in a kernel (#3891)

  Show description for d02e98c
  slaren

  authored
  d02e98c

  Copy full SHA for d02e98c

• llama : implement YaRN RoPE scaling (#2268)

  Show description for 898aeca
  

  cebtenzzre

  and

  jquesnelle

  authored
  898aeca

  Copy full SHA for 898aeca

• llm : fix llm_build_kqv taking unused tensor (benign, #3837)

  ggerganov

  committed
  c43c2da

  Copy full SHA for c43c2da

• llm : fix falcon norm after refactoring (#3837)

  ggerganov

  committed
  523e49b

  Copy full SHA for 523e49b

• metal : multi-simd softmax (#3710)

  Show description for e16b9fa
  ggerganov

  authored
  e16b9fa

  Copy full SHA for e16b9fa

• common : minor (#3715)

  ggerganov

  committed
  ff8f9a8

  Copy full SHA for ff8f9a8

Pagination

Previous

Next