ggerganov/llama.cpp

Pull Requests Commits

chat : fix build on arm64

ngxson committed 11 hours ago

75cb7b56

ggml : refactor forward_dup for cpu backend (#16062)

ngxson committed 11 hours ago

Verified 0dd58b68

ggml-amx : fix ggml_amx_init() on generic Linux (#16049)

angt committed 19 hours ago

Verified 69ffd891

cmake : fix static linking for OpenMP on Unix-like systems (#16031)

angt committed 19 hours ago

Verified 246c0d9c

opencl: optimize mxfp4 kernels (#16037)

shawngu-quic committed 21 hours ago

Verified 3edd87cd

rename optimize_graph to graph_optimize (#16082)

jeffbolznv committed 21 hours ago

Verified c0b45097

CUDA: Optimize PAD_REFLECT_1D (#15957)

bugparty committed 21 hours ago

Verified 38dbdf4c

CUDA: fix compilation on CC 6.0 (#16091)

JohannesGaessler committed 22 hours ago

Verified 368560a1

Add resumable downloads for llama-server model loading (#15963)

ericcurtin committed 1 day ago

Verified 4ca088b0

metal : use function constants for mul_mv_ext kernels (#16074)

ggerganov committed 1 day ago

Verified 703f9e32

cuda : add missing F32<->I32 entries in ggml_cuda_cpy_fn (#16060)

CISC committed 1 day ago

Verified ad6bd908

server : include usage statistics only when user request them (#16052)

rgerganov committed 1 day ago

Verified 2b6b55a5

llama : bump max seq limit from 64 to 256 (#15916)

ggerganov committed 1 day ago

Verified e58174ce

metal : improve F32, F16 and BF16 mat-vec multiplication (#16057)

ggerganov committed 1 day ago

Verified b213fce8

metal : avoid call free for non-owned buffer (#16067)

jhen0409 committed 1 day ago

Verified e00f3fd8

metal : handle nil cv during pipeline creation (#16065)

ggerganov committed 1 day ago

Verified f2f28380

CANN: Remove print (#16044)

noemotiovon committed 1 day ago

Verified 62c3b645

GGML WebGPU: Support for ADD, MUL, RMS_NORM, GET_ROWS operators (#16018)

reeselevine committed 1 day ago

Verified d304f459

metal : refactor + optimize v2 (#15995)

ggerganov committed 1 day ago

Verified 0320ac52

SvelteKit-based WebUI (#14839)

allozaur committed 1 day ago

Verified a7a98e0f

convert : add Llama4ForCausalLM (#16042)

ngxson committed 1 day ago

Verified 8f8f2274

CUDA: fix FA occupancy, optimize tile kernel (#15982)

JohannesGaessler committed 2 days ago

Verified c959b676

common : Fix corrupted memory error on json grammar initialization (#16038)

dralves committed 2 days ago

Verified cd08fc3e

vulkan: automatically remove unsupported devices (#15976)

netrunnereve committed 2 days ago

Verified cb5bb6cc

ci : revert back to macos-13 for macOS-latest-cmake-x64 (#16040)

danbev committed 2 days ago

Verified a91d035b

llama-quant : fix the verification of attention layers for encoder-decoder models (#16023)

DamonFool committed 2 days ago

Verified 745cbcf2

examples : support encoder-decoder models in the simple example (#16002)

DamonFool committed 2 days ago

Verified 1cbd80f8

model : add OLMo3 support (#16015)

2015aroras committed 2 days ago

Verified 85286f35

CANN: Optimize ggml_cann_set_device (#15935)

noemotiovon committed 2 days ago

Verified d5fabe36

llama-bench: add --n-cpu-moe support (#15952)

jacekpoplawski committed 3 days ago

Verified 8ff20609

Older