Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 28, 2024

• sync : ggml

  ggerganov

  committed
  3168dbf

  Copy full SHA for 3168dbf

• sync : llama.cpp (ggml/0)

  ggerganov

  committed
  1711bb3

  Copy full SHA for 1711bb3

• ggml : make i-quants work with super-blocks of 64 (CPU,Metal) (llama/5760)

  Show description for 2533305
  

  2 peopleauthored and

  ggerganov

  committed
  2533305

  Copy full SHA for 2533305

• Attempt to fix android build (llama/5752)

  Show description for 0eca512
  

  2 peopleauthored and

  ggerganov

  committed
  0eca512

  Copy full SHA for 0eca512

• IQ4_XS: a 4.25 bpw quantization (llama/5747)

  Show description for 013e394
  

  2 peopleauthored and

  ggerganov

  committed
  013e394

  Copy full SHA for 013e394

• cuda : replace remaining shfl_xor with calls to warp_reduce functions (llama/5744)

  

  Engininja2

  authored and

  ggerganov

  committed
  d83f371

  Copy full SHA for d83f371

• ggml-quants : fix avx2 iq1_s vec_dot when compiled with gcc (llama/5742)

  

  Engininja2

  authored and

  ggerganov

  committed
  1c71816

  Copy full SHA for 1c71816

• Adding IQ2_S and IQ2_M to complete coverage of the 2-3 bit quantization range (llama/5721)

  Show description for 7b1d8ea
  

  3 peoplecommitted
  7b1d8ea

  Copy full SHA for 7b1d8ea

• CUDA: fix DEBUG_CUDA_MALLOC (llama/5729)

  

  JohannesGaessler

  authored and

  ggerganov

  committed
  b1f7223

  Copy full SHA for b1f7223

• Add support for soft_max ALiBi (llama/5639)

  Show description for 8408a4b
  

  2 peopleauthored and

  ggerganov

  committed
  8408a4b

  Copy full SHA for 8408a4b

• ggml-quants : provide ggml_vqtbl1q_u8 for 64bit compatibility (llama/5711)

  Show description for 72849c2
  

  rgryta

  authored and

  ggerganov

  committed
  72849c2

  Copy full SHA for 72849c2

• add google magika inference example (ggml/748)

  Show description for c19c28b
  

  slaren

  authored and

  ggerganov

  committed
  c19c28b

  Copy full SHA for c19c28b

Commits on Feb 26, 2024

• stream.wasm : fix invalid memory access when no segments (ggerganov#1902)

  Show description for 0d8fd84
  Andrews54757

  authored
  0d8fd84

  Copy full SHA for 0d8fd84

Commits on Feb 25, 2024

• talk-llama : sync llama.cpp

  ggerganov

  committed
  3170841

  Copy full SHA for 3170841

• sync : ggml

  ggerganov

  committed
  7a6e385

  Copy full SHA for 7a6e385

• sync : llama.cpp (ggml/0)

  ggerganov

  committed
  578e47e

  Copy full SHA for 578e47e

• code : normalize enum names (llama/5697)

  Show description for fac5b43
  ggerganov

  committed
  fac5b43

  Copy full SHA for fac5b43

• IQ3_S: a much better alternative to Q3_K (llama/5676)

  Show description for 9e7c521
  

  2 peopleauthored and

  ggerganov

  committed
  9e7c521

  Copy full SHA for 9e7c521

• Introduce backend GUIDs (ggml/743)

  Show description for 1cb64f7
  

  2 peopleauthored and

  ggerganov

  committed
  1cb64f7

  Copy full SHA for 1cb64f7

Commits on Feb 24, 2024

• talk, talk-llama : pass text_to_speak as a file (ggerganov#1865)

  Show description for f18738f
  tamo

  authored
  f18738f

  Copy full SHA for f18738f

Commits on Feb 23, 2024

• whisper : add SYCL support (ggerganov#1863)

  Show description for a0ddd83
  abhilash1910

  authored
  a0ddd83

  Copy full SHA for a0ddd83

Commits on Feb 22, 2024

• talk-llama : sync llama.cpp

  ggerganov

  committed
  a250690

  Copy full SHA for a250690

• sync : ggml

  ggerganov

  committed
  7b1ff21

  Copy full SHA for 7b1ff21

• ggml : always define ggml_fp16_t as uint16_t (llama/5666)

  Show description for e5d06cf
  ggerganov

  committed
  e5d06cf

  Copy full SHA for e5d06cf

• ci : fix whitespace

  ggerganov

  committed
  31891db

  Copy full SHA for 31891db

• ggml : 32-bit arm compat (ggerganov#1891)

  Show description for 5fdb27f
  ggerganov

  authored
  5fdb27f

  Copy full SHA for 5fdb27f

• sync : ggml

  ggerganov

  committed
  6b16927

  Copy full SHA for 6b16927

• sync : llama.cpp (ggml/0)

  Show description for ce41149
  ggerganov

  committed
  ce41149

  Copy full SHA for ce41149

• conext add name (llama/5624)

  Show description for 208de95
  

  airMeng

  authored and

  ggerganov

  committed
  208de95

  Copy full SHA for 208de95

• Update ggml_sycl_op_mul_mat_vec_q (llama/5502)

  Show description for c2ce39c
  

  2 peopleauthored and

  ggerganov

  committed
  c2ce39c

  Copy full SHA for c2ce39c

• Refactor validation and enumeration platform checks into functions to clean up ggml_vk_instance_init()

  

  0cc4m

  authored and

  ggerganov

  committed
  8daa534

  Copy full SHA for 8daa534

• Add check for VK_KHR_portability_enumeration for MoltenVK support

  

  0cc4m

  authored and

  ggerganov

  committed
  9fca69b

  Copy full SHA for 9fca69b

• Add preprocessor checks for Apple devices.

  Show description for b26c645
  

  dokterbob

  authored and

  ggerganov

  committed
  b26c645

  Copy full SHA for b26c645

• Resolve ErrorIncompatibleDriver with Vulkan on MacOS.

  Show description for 1879ec5
  

  dokterbob

  authored and

  ggerganov

  committed
  1879ec5

  Copy full SHA for 1879ec5

• Allow for Vulkan build with Accelerate.

  Show description for c6e53cf
  

  dokterbob

  authored and

  ggerganov

  committed
  c6e53cf

  Copy full SHA for c6e53cf

Pagination

Previous

Next