Commits

Branch selector

 master

User selector

ggerganov

Datepicker

All time

Commit History

Commits on Oct 7, 2024

• release : v1.7.1

  ggerganov

  committed
  ebca09a

Commits on Oct 5, 2024

• release : v1.7.0

  ggerganov

  committed
  6a94163

• scripts : bench v3-turbo

  ggerganov

  committed
  8a35b58

• whisper : remove mel leftover constants (396089f)

  ggerganov

  committed
  1789abc

• whisper : zero-out the KV cache upon clear (ggerganov#2445)

  ggerganov

  committed
  847f94f

• objc : fix build

  ggerganov

  committed
  6e40108

• metal : zero-init buffer contexts (#0)

  ggerganov

  committed
  1ba185f

• whisper : revert mel-related changes (#0)

  Show description for 396089f
  ggerganov

  committed
  396089f

• whisper : adapt to latest ggml (skip) (#0)

  ggerganov

  committed
  9419124

• scripts : sync ggml-backend.cpp

  ggerganov

  committed
  f7d55e0

• whisper : fix excessive memory usage (ggerganov#2443)

  Show description for f62a546
  ggerganov

  authored
  f62a546

Commits on Oct 3, 2024

• talk-llama : sync llama.cpp

  ggerganov

  committed
  ccc2547

• metal : reduce command encoding overhead (llama/9698)

  ggerganov

  committed
  162a455

• sync : ggml

  ggerganov

  committed
  ff2cb08

• ggml : define missing HWCAP flags (llama/9684)

  Show description for 6c91da8
  

  ggerganov

  and

  wtarreau

  committed
  6c91da8

• ggml : fix GGML_MAX_N_THREADS + improve formatting (ggml/969)

  ggerganov

  committed
  5963004

Commits on Oct 1, 2024

• whisper : add large-v3-turbo (ggerganov#2440)

  ggerganov

  authored
  2ef717b

Commits on Sep 27, 2024

• tests : remove test-backend-ops (ggerganov#2434)

  ggerganov

  committed
  8feb375

Commits on Sep 25, 2024

• ci : disable failing CUDA and Java builds

  ggerganov

  committed
  69339af

Commits on Sep 24, 2024

• make : remove "talk" target until updated

  ggerganov

  committed
  451e9ee

• ggml : add ggml-cpu-impl.h (skip) (#0)

  ggerganov

  committed
  1133ac9

• sync : ggml

  ggerganov

  committed
  76d27ee

• talk-llama : sync llama.cpp

  ggerganov

  committed
  fe18c29

• log : add CONT level for continuing previous log entry (llama/9610)

  ggerganov

  committed
  3b183cf

• metal : use F32 prec for K*Q in vec FA (llama/9595)

  Show description for 896c41e
  ggerganov

  committed
  896c41e

• examples : adapt to ggml.h changes (ggml/0)

  Show description for 54e5095
  ggerganov

  committed
  54e5095

• ggml : refactoring (llama/#0)

  Show description for 3429109
  ggerganov

  committed
  3429109

• ggml : fix builds (llama/0)

  Show description for d245d7a
  ggerganov

  committed
  d245d7a

• ggml : fix trailing whitespace (llama/0)

  Show description for d661283
  ggerganov

  committed
  d661283

• cmake : do not hide GGML options + rename option (llama/9465)

  Show description for 1fd7899
  ggerganov

  committed
  1fd7899

• metal : handle zero-sized allocs (llama/9466)

  ggerganov

  committed
  a2cb5b4

• common : reimplement logging (llama/9418)

  Show description for 288ae51
  ggerganov

  committed
  288ae51

• cmake : use list(APPEND ...) instead of set() + dedup linker (llama/9463)

  Show description for 66b00fa
  

  ggerganov

  and

  Xarbirus

  committed
  66b00fa

• metal : fix compile warning with GGML_METAL_NDEBUG (llama/0)

  ggerganov

  committed
  a785232

• cuda : fix FA Q src index (1 -> 0) (llama/9374)

  ggerganov

  committed
  26225f1

Pagination

PreviousNext