ggml-org/llama.cpp

Pull Requests Commits

server: add --models-memory-max parameter to allow dynamically unloading models when they exceed a memory size threshold

0cc4m committed 38 days ago

e4d2e198

common : remove unused regex-partial (#25118)

o7si committed 38 days ago

Verified 277a105d

jinja, chat: add --reasoning-preserve flag (#25105)

ngxson committed 38 days ago

Verified b3fed31b

Revert "ui: fix accessibility for hover-gated interactive elements assisted by claude(in debugging and tests) (#24727)" (#25098)

allozaur committed 39 days ago

Verified dbdaece2

ui: fix stop and reasoning skip in single-model mode (#25084)

ServeurpersoCom committed 39 days ago

Verified 7cb8576e

dflash: refactor draft model conversion (#25110)

ruixiang63 committed 39 days ago

Verified fa72bc68

chat : implement minicpm5 parser (#24889)

aldehir committed 39 days ago

Verified c818263f

jinja: add --dump-prog for debugging (#25086)

ngxson committed 39 days ago

Verified f68a788b

spec : add DFlash support (#22105)

ruixiang63 committed 39 days ago

Verified d1b34251

common : allow --offline in llama download (#25091)

angt committed 39 days ago

Verified c1a1c8ee

logs : reduce v2 (#25078)

ggerganov committed 39 days ago

Verified 27c8bb4f

opencl: flash attention improvement (#25069)

wanghqc committed 39 days ago

Verified ebd048fc

[CUDA] Added a cudaMemcpy2DAsync fast path to ggml_cuda_cpy (#25057)

gaugarg-nv committed 40 days ago

Verified 0ed235ea

sycl : fix failed ut cases of norm (#25044)

arthw committed 40 days ago

Verified 9bebfcb4

vulkan: fix step operator for 0 input (#25036)

0cc4m committed 40 days ago

Verified 0b6529d8

binaries : Improve rpc-server and export-graph-ops names. (#25045)

ckastner committed 40 days ago

Verified c299a92c

ci : add windows-openvino to check-release (#25022)

CISC committed 40 days ago

Verified 0275c0f8

tests : fix test-chat-template --no-common option (#25075)

CISC committed 40 days ago

Verified 83d385b4

app : allow --version, --licenses & --help (#25054)

angt committed 40 days ago

Verified 050ee92d

sched : reintroduce less synchronizations during split compute (#20793)

aendk committed 41 days ago

Verified 3fc4e105

devops : add llama in all docker images (#25035)

angt committed 41 days ago

Verified 5d8ccdf9

arg: fix handling --spec-draft-hf and --hf-repo-v (#25043)

ngxson committed 41 days ago

Verified 024930c6

openvino: Update to OV 2026.2.1, self-contained release packages, operator improvements (#24974)

ravi9 committed 41 days ago

Verified 5397c361

ggerganov committed 41 days ago

e7ea94af

ggml : bump version to 0.15.3 (ggml/1550)

ggerganov committed 41 days ago

96183e98

vulkan: opt mul_mat_vecq for mi50 (#22933)

chraac committed 41 days ago

Verified 487a6cc1

vulkan: add INTEL_XE1 arch enum and enable coopmat1 on Intel Xe-LPG Plus (#24404)

fish-jiang committed 41 days ago

Verified 5a6a0dd7

ui: fix accessibility for hover-gated interactive elements assisted by claude(in debugging and tests) (#24727)

sanjayahari committed 41 days ago

Verified ded1561b

vulkan: Workaround compiler bug in conv2d coopmat2 path (#24924)

jeffbolznv committed 41 days ago

Verified 9df06805

CUDA: add cublasSgemmBatched mapping for HIP/MUSA vendor headers (#25033)

leonardHONG committed 41 days ago

Verified 2f18fe13

Older