Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 1, 2025

• convert : remove redundant code (#15708)

  Show description for 4b20d8b
  DamonFool

  authored
  4b20d8b

  Copy full SHA for 4b20d8b

• Vulkan: Add Integer Dot Product mul_mat_vec shader for legacy quants (#14903)

  Show description for 02c1813
  0cc4m

  authored
  02c1813

  Copy full SHA for 02c1813

• ggml : WebGPU add TRANSPOSE and RESHAPE to supported ops (#15695)

  Show description for 77dee9d
  

  3 peopleauthored
  77dee9d

  Copy full SHA for 77dee9d

• docs : add Hunyuan to models section (#15707)

  Show description for 4795c91
  DamonFool

  authored
  4795c91

  Copy full SHA for 4795c91

• CUDA: fix build error from ambiguous __half conversions in conv2d (#15690)

  Show description for b66df9d
  qnixsynapse

  authored
  b66df9d

  Copy full SHA for b66df9d

• CANN: Optimize MUL_MAT_ID (#15658)

  hipudding

  authored
  b9382c3

  Copy full SHA for b9382c3

• CANN: fix RoPE cache issue on multi-device (#15629)

  Show description for 3dc7397
  hipudding

  authored
  3dc7397

  Copy full SHA for 3dc7397

Commits on Aug 31, 2025

• sampling : optimize samplers by reusing bucket sort (#15665)

  Show description for e92d53b
  

  ggerganov

  and

  JohannesGaessler

  authored
  e92d53b

  Copy full SHA for e92d53b

• server : enable /slots by default and make it secure (#15630)

  Show description for 0d161f0
  ggerganov

  authored
  0d161f0

  Copy full SHA for 0d161f0

• metal : fix checks for available FA kernels (#15700)

  Show description for 4efd5a8
  ggerganov

  authored
  4efd5a8

  Copy full SHA for 4efd5a8

• llama : fix fattn reserve call n_seqs parameter (#15699)

  Show description for 2749662
  slaren

  authored
  2749662

  Copy full SHA for 2749662

• llama : separate compute buffer reserve from fattn check (#15696)

  Show description for 9777032
  slaren

  authored
  9777032

  Copy full SHA for 9777032

• ci : explicitly set fa off or on (#15692)

  CISC

  authored
  7d3c9f2

  Copy full SHA for 7d3c9f2

• vulkan: handle large sizes for get_rows (#15686)

  jeffbolznv

  authored
  bbbf5ec

  Copy full SHA for bbbf5ec

• vulkan: mul_mat_id coopmat2 optimizations (#15546)

  Show description for c37052a
  jeffbolznv

  authored
  c37052a

  Copy full SHA for c37052a

• vulkan : remove unused portability_enumeration_ext variable (#15679)

  Show description for 5c16b9c
  danbev

  authored
  5c16b9c

  Copy full SHA for 5c16b9c

• vulkan: Allow fallback to sysmem memory when vidmem is full (#15649)

  Show description for b97c9ed
  jeffbolznv

  authored
  b97c9ed

  Copy full SHA for b97c9ed

• vulkan: clamp matmul and FA results to the max finite value (#15652)

  Show description for 94e82c7
  jeffbolznv

  authored
  94e82c7

  Copy full SHA for 94e82c7

Commits on Aug 30, 2025

• ggml: update kleidiai to v1.13.0 (#15663)

  chaxu01

  authored
  4d74393

  Copy full SHA for 4d74393

• Update build.md to remove MSVC arm64 notes (#15684)

  Show description for dd89255
  slaren

  authored
  dd89255

  Copy full SHA for dd89255

• llama: use FA + max. GPU layers by default (#15434)

  Show description for e81b8e4
  JohannesGaessler

  authored
  e81b8e4

  Copy full SHA for e81b8e4

• CUDA: use FP32 arithmetic for conv2d (#15683)

  JohannesGaessler

  authored
  38ad381

  Copy full SHA for 38ad381

• vulkan: Skip syncing for prealloc_y when it is reused (#15544)

  jeffbolznv

  authored
  696fccf

  Copy full SHA for 696fccf

• CANN: FIx compiler warnings (#15661)

  Show description for ef47691
  noemotiovon

  authored
  ef47691

  Copy full SHA for ef47691

Commits on Aug 29, 2025

• server : removed obsolete doc (#15670)

  Show description for d82f6aa
  l29ah

  authored
  d82f6aa

  Copy full SHA for d82f6aa

• scripts: strip "AMD Instinct" from GPU name (#15668)

  JohannesGaessler

  authored
  3d16b29

  Copy full SHA for 3d16b29

• server : add documentation for `parallel_tool_calls` param (#15647)

  Show description for 792b44f
  

  ExtReMLapin

  and

  Pierre F

  authored
  792b44f

  Copy full SHA for 792b44f

• CUDA: fix bug in rms_norm fusion (#15660)

  Show description for 8101786
  am17an

  authored
  8101786

  Copy full SHA for 8101786

• chat : Seed OSS thinking + tool call support (#15552)

  Show description for 60e5eee
  

  pwilkin

  and

  CISC

  authored
  60e5eee

  Copy full SHA for 60e5eee

• CUDA: fuse adds, fuse add with rms norm (#15631)

  Show description for 009b709
  am17an

  authored
  009b709

  Copy full SHA for 009b709

• nvidia nemotron nano v2 (nemotronh) (#15507)

  Show description for e8d99dd
  

  3 peopleauthored
  e8d99dd

  Copy full SHA for e8d99dd

Commits on Aug 28, 2025

• fix: Compute the full sum in llama-eval-callback, not just the sum of printed values (#15637)

  Show description for a8bca68
  gabe-l-hart

  authored
  a8bca68

  Copy full SHA for a8bca68

• CUDA: add conv2d (#15635)

  Show description for c97dc09
  mnehete32

  authored
  c97dc09

  Copy full SHA for c97dc09

• ggml-cpu: fix invalid hsum build in debug s390x (#15634)

  Show description for 6c442f4
  taronaeo

  authored
  6c442f4

  Copy full SHA for 6c442f4

• ggml : fix SSM_SCAN for n_groups > 1 (#15625)

  compilade

  authored
  7380414

  Copy full SHA for 7380414

Pagination

Previous

Next