ggerganov/llama.cpp

Pull Requests Commits

Specify podman works in Container documentation

ericcurtin committed 247 days ago

947158ee

metal : adjust support conditions for norm operators (#11671)

ggerganov committed 247 days ago

Verified d774ab3a

CUDA: support for mat. mul. with ne03 != ne13 (#11656)

JohannesGaessler committed 247 days ago

Verified fa62da9b

llava: add quantization for the visual projector LLAVA, Qwen2VL (#11644)

samkoesnadi committed 247 days ago

Verified 1ec20808

`sync`: minja (#11641)

ochafik committed 247 days ago

Verified 9f4cc8f8

CUDA: non-contiguous (RMS) norm support (#11659)

JohannesGaessler committed 247 days ago

Verified fd08255d

HIP: force max threads per block to be 1024 (#11621)

fxzjshm committed 247 days ago

Verified 3ec9fd4b

server : add try..catch to places not covered by set_exception_handler (#11620)

ngxson committed 247 days ago

Verified 3962fc1a

arg : list RPC devices first when using --list-devices (#11655)

rgerganov committed 247 days ago

Verified 1bef571f

`tool-call`: command r7b fix for normal responses (#11608)

ochafik committed 247 days ago

Verified db288b60

readme : add llm_client Rust crate to readme bindings (#11628)

ShelbyJenkins committed 248 days ago

Verified 106045e7

swift : fix llama-vocab api usage (#11645)

jhen0409 committed 248 days ago

Verified f117d84b

metal : use residency set for other platforms (#11648)

jhen0409 committed 248 days ago

Verified 534c46b5

authors : update

ggerganov committed 248 days ago

Verified 387a1598

ggerganov committed 248 days ago

Verified 7c9e0ca5

cmake: Add ability to pass in GGML_BUILD_NUMBER (ggml/1096)

ckastner committed 248 days ago

Verified 8f8290ad

ci : do not stale-close roadmap issues

ggerganov committed 248 days ago

Verified b34aedd5

`tool-call`: allow `--chat-template chatml` w/ `--jinja`, default to chatml upon parsing issue, avoid double bos (#11616)

ochafik committed 248 days ago

Verified cde38332

server : (webui) revert hacky solution from #11626 (#11634)

ngxson committed 248 days ago

Verified b3451785

server : (webui) allow typing and submitting during llm response (#11626)

woof-dog committed 248 days ago

Verified 1d1e6a90

server : remove CPPHTTPLIB_NO_EXCEPTIONS define (#11622)

danbev committed 248 days ago

Verified 5598f475

ggerganov committed 249 days ago

Verified 8ec05832

CUDA: fix Volta FlashAttention logic (#11615)

JohannesGaessler committed 249 days ago

Verified 21c84b5d

server : (webui) Fix Shift+Enter handling (#11609)

mashdragon committed 249 days ago

Verified d92cb67e

HIP: fix flash_attn_stream_k_fixup warning (#11604)

JohannesGaessler committed 249 days ago

Verified 6eecde3c

CUDA/HIP: add support for selectable warp size to mmv (#11519)

IMbackK committed 249 days ago

Verified 396856b4

HIP: add GGML_CUDA_CC_IS_* for amd familys as increasing cc archtectures for amd gpus are not supersets of eatch other (#11601)

IMbackK committed 249 days ago

Verified 4d0598e1

nit: more informative crash when grammar sampler fails (#11593)

ochafik committed 249 days ago

Verified 90f9b88a

CUDA: use mma PTX instructions for FlashAttention (#11583)

JohannesGaessler committed 249 days ago

Verified 864a0b67

Name colors (#11573)

ericcurtin committed 249 days ago

Verified 84ec8a58

Older