ngxson/llama.cpp

Pull Requests Commits

mtmd: fix --no-warmup

ngxson committed 13 days ago

2b675878

CANN: Disable Ger operator of OUT_PROD on 310p device (#17563)

TianHao324 committed 13 days ago

Verified a2b0fe8d

ggml : remove redundant n_copies check when setting input/output (#17612)

danbev committed 13 days ago

Verified 7f3a72a8

codeowners : remove ericcurtin (#17658)

ericcurtin committed 13 days ago

Verified b9a37717

llama : fix signed comparison warning on FreeBSD (#17497)

angt committed 13 days ago

Verified f3a9674a

convert: add error message for mistral3 quantized weight (#17686)

ngxson committed 13 days ago

Verified 2c453c6c

server: remove default "gpt-3.5-turbo" model name (#17668)

ngxson committed 13 days ago

Verified 5d6bd842

server: fixing naming conflict res_error in server-models.cpp (#17679)

w169q169 committed 13 days ago

Verified fd3abe84

server: explicitly set exec path when create new instance (#17669)

ngxson committed 13 days ago

Verified 682e6658

ci : skip winget update when not in ggml-org (#17465)

angt committed 13 days ago

Verified 4574f294

ggml : add fallback definition for HWCAP2_SVE2 (#17683)

angt committed 13 days ago

Verified ab6726ee

Add context info to server error (#17663)

allozaur committed 13 days ago

Verified cee92af5

ggml-cuda: reorder only relevant nodes (#17639)

am17an committed 14 days ago

Verified ed320899

release: fix duplicate libs, store symbolic links (#17299)

taronaeo committed 14 days ago

Verified 7b6d7453

enhance argsort for UT (#17573)

NeoZhangJianyu committed 14 days ago

Verified 98bd9ab1

Override SSM_A op for Qwen3 Next to reduce splits (#17587)

pwilkin committed 14 days ago

Verified 746f9ee8

ops.md: update vulkan support (#17661)

jeffbolznv committed 14 days ago

Verified 9810cb82

mtmd: add mtmd_context_params::warmup option (#17652)

ngxson committed 14 days ago

Verified ecf74a84

fix: llama arch implementation (#17665)

giladgd committed 14 days ago

Verified 00c361fe

server: introduce API for serving / loading / unloading multiple models (#17470)

ngxson committed 14 days ago

Verified ec18edfc

common: improve verbosity level definitions (#17630)

ngxson committed 14 days ago

Verified 77334097

model: support Ministral3 (#17644)

ngxson committed 14 days ago

Verified cd3c1189

metal : add FA head size 48 (#17619)

ggerganov committed 14 days ago

Verified 649495c9

ggml : extend the GGML_SCHED_NO_REALLOC debug logic of the scheduler (#17617)

ggerganov committed 14 days ago

Verified 90c72a61

llama-graph: avoid expand_forward for fusion (#17633)

am17an committed 14 days ago

Verified 6eea6669

contributing: update guidelines for AI-generated code (#17625)

ngxson committed 15 days ago

Verified ff90508d

cmake : add option to build and link LibreSSL (#17552)

angt committed 15 days ago

Verified 0a4aeb92

model: LFM2-VL fixes (#17577)

tdakhran committed 15 days ago

Verified 2ba71951

clip: fix nb calculation for qwen3-vl (#17594)

ngxson committed 15 days ago

Verified 7f8ef50c

cli: add migration warning (#17620)

ngxson committed 15 days ago

Verified 3c136b21

Older