ggerganov/llama.cpp

Pull Requests Commits

llama : fix sanity checks during quantization

ggerganov committed 61 days ago

Verified 01c9e9fd

ci : move release details to the top visible by default (#17719)

CISC committed 61 days ago

Verified b3e3060f

ggml, llama : use defaulted constructors/destructors (#17649)

GermanAizek committed 61 days ago

Verified 37adc9c6

build: document how to compile with Vulkan using Debian/Ubuntu packages (#17688)

socram8888 committed 61 days ago

Verified 16cc3c60

server: add --media-path for local media files (#17697)

ngxson committed 61 days ago

Verified 13628d8b

mtmd: fix --no-warmup (#17695)

ngxson committed 61 days ago

Verified a96283ad

ci : RVV1.0 builds with tests (#16682)

alitariq4589 committed 61 days ago

Verified 4eba8d94

vulkan: Reduce temporary memory usage for TOP_K (#17623)

jeffbolznv committed 61 days ago

Verified 61bde8e2

cmake : add utf8 compilation options for msvc (#17682)

xiaobing318 committed 61 days ago

Verified e251e5eb

Server: Change Invalid Schema from Server Error (500) to User Error (400) (#17572)

chadvoegele committed 62 days ago

Verified c4357dcc

ggml : use svcntb() for SVE vector length detection (#17474)

angt committed 62 days ago

Verified e148380c

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

TianHao324 committed 62 days ago

Verified a2b0fe8d

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

danbev committed 62 days ago

Verified 7f3a72a8

codeowners : remove ericcurtin (#17658)

ericcurtin committed 62 days ago

Verified b9a37717

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

angt committed 62 days ago

Verified f3a9674a

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

ngxson committed 62 days ago

Verified 2c453c6c

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

ngxson committed 62 days ago

Verified 5d6bd842

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

w169q169 committed 62 days ago

Verified fd3abe84

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

ngxson committed 62 days ago

Verified 682e6658

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

angt committed 62 days ago

Verified 4574f294

ggml : add fallback definition for HWCAP2_SVE2 (#17683)

angt committed 62 days ago

Verified ab6726ee

Add context info to server error (#17663)

allozaur committed 62 days ago

Verified cee92af5

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

am17an committed 62 days ago

Verified ed320899

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

taronaeo committed 62 days ago

Verified 7b6d7453

enhance argsort for UT (#17573)

NeoZhangJianyu committed 62 days ago

Verified 98bd9ab1

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

pwilkin committed 62 days ago

Verified 746f9ee8

ops.md: update vulkan support (#17661)

jeffbolznv committed 62 days ago

Verified 9810cb82

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

ngxson committed 62 days ago

Verified ecf74a84

fix: llama arch implementation (#17665)

giladgd committed 62 days ago

Verified 00c361fe

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

ngxson committed 62 days ago

Verified ec18edfc

Older