Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 16, 2024

• ggml : fix and optimize ppc64le (ggml/849)

  Show description for 5cd42ee
  

  penghongbo

  authored and

  ggerganov

  committed
  5cd42ee

• ggml : remove duplicate include of ggml-common.h (ggml/853)

  Show description for ee718f3
  

  danbev

  authored and

  ggerganov

  committed
  ee718f3

• ggml : fix loongson compile warnings (llama/7537)

  Show description for 9a16c64
  

  ggerganov

  and

  junchao-loongson

  committed
  9a16c64

• ggml : fix loongarch build (O2 issue) (llama/7636)

  

  junchao-loongson

  authored and

  ggerganov

  committed
  29cfeef

• ggml: aarch64: SVE kernels for q8_0_q8_0, q4_0_q8_0 vector dot (llama/7433)

  Show description for 37ed71c
  

  msy-kato

  authored and

  ggerganov

  committed
  37ed71c

• ggml : silence UB sanitizer error during iq2_xxs quantization (llama/0)

  ggerganov

  committed
  8cd7a3d

• ggml : drop support for QK_K=64 (llama/7473)

  Show description for 45ddda8
  ggerganov

  committed
  45ddda8

• ggml : add loongarch lsx and lasx support (llama/6454)

  Show description for eb26f55
  

  2 peopleauthored and

  ggerganov

  committed
  eb26f55

• ggml : fix another case of quants nans (llama/7387)

  

  slaren

  authored and

  ggerganov

  committed
  bf5fc81

• ggml : fix quants nans when all the group weights are very close to zero (llama/7313)

  

  slaren

  authored and

  ggerganov

  committed
  2bc6483

• ggml-quants, llama : removed excess checks (llama/7274)

  

  GermanAizek

  authored and

  ggerganov

  committed
  c00149c

• Add support for properly optimized Windows ARM64 builds with LLVM and MSVC (llama/7191)

  Show description for 488ad99
  

  max-krasnyansky

  and

  ggerganov

  committed
  488ad99

Commits on May 13, 2024

• ggml : try fix ppc64 (#0)

  ggerganov

  committed
  9506267

• ggml : optimize for ppc64le using VSX intrinsics (ggml/784)

  Show description for 40aeeee
  

  penghongbo

  and

  ggerganov

  committed
  40aeeee

• ggml : introduce bfloat16 support (llama/6412)

  Show description for a2ad810
  

  jart

  authored and

  ggerganov

  committed
  a2ad810

• add basic tensor data validation function (llama/6884)

  Show description for 9d4c8b8
  

  slaren

  authored and

  ggerganov

  committed
  9d4c8b8

• ggml : fix MIN / MAX macros (llama/6904)

  Show description for a15fb5c
  ggerganov

  committed
  a15fb5c

• ggml : move 32-bit arm compat in ggml-impl.h (llama/6865)

  Show description for 63fd148
  ggerganov

  committed
  63fd148

• ggml : add llamafile sgemm (llama/6414)

  Show description for 7a4f7d8
  

  jart

  authored and

  ggerganov

  committed
  7a4f7d8

• build: fix and ignore msvc warnings (ggml/805)

  

  iboB

  authored and

  ggerganov

  committed
  8dcefdf

Commits on Apr 9, 2024

• llama : add Command R Plus support (llama/6491)

  Show description for 5263328
  

  5 peoplecommitted
  5263328

Commits on Mar 27, 2024

• sync : ggml (ggerganov#2001)

  Show description for 2948c74
  ggerganov

  authored
  2948c74

Commits on Mar 15, 2024

• ggml : reuse quantum structs across backends (llama/5943)

  Show description for 46e3c3f
  ggerganov

  committed
  46e3c3f

• ggml : fix UB in IQ2_S and IQ3_S (llama/6012)

  ggerganov

  committed
  ef24ae0

• 1.5 bit: we can do even better (llama/5999)

  Show description for 9dc60fc
  

  2 peopleauthored and

  ggerganov

  committed
  9dc60fc

• ggml, ci : Windows ARM runner and build fixes (llama/5979)

  Show description for d73a636
  

  Xarbirus

  authored and

  ggerganov

  committed
  d73a636

• Better 1.5 bit quantization (llama/5971)

  Show description for f79d0d4
  

  2 peopleauthored and

  ggerganov

  committed
  f79d0d4

• ggml : fix unnecessary f32 -> f16 -> f32 casts (mmla) (llama/5951)

  ggerganov

  committed
  653d2e8

• ggml : remove old quantization functions (llama/5942)

  Show description for 2fef660
  ggerganov

  committed
  2fef660

• ggml : add ggml-common.h to deduplicate shared code (llama/5940)

  Show description for 24eba5a
  ggerganov

  committed
  24eba5a

Commits on Mar 8, 2024

• ggml : try fix 32-bit arm compat (ggerganov#1938)

  Show description for ce945b5
  ggerganov

  authored
  ce945b5

• ggml : use uint8x16_t return type for ggml_vqtbl1q_u8 (llama/5894)

  Show description for 2f6f1d4
  

  bobqianic

  authored and

  ggerganov

  committed
  2f6f1d4

• quants : use MM256_SET_M128I consistently to fix gcc 7 build (llama/5889)

  

  cebtenzzre

  authored and

  ggerganov

  committed
  8edfc54

• ggml : fix IQ3_S AVX implementation (llama/5834)

  Show description for bd26876
  ggerganov

  committed
  bd26876

• ggml : IQ3_S improvements (llama/5829)

  Show description for 21d2951
  

  2 peopleauthored and

  ggerganov

  committed
  21d2951

Pagination

PreviousNext