ngxson/llama.cpp

Pull Requests Commits

mtmd: explicitly forbidden inclusion of private header and libcommon

ngxson committed 60 days ago

dbd5144f

Add a search field on model selector / improve mobile display (#17765)

ServeurpersoCom committed 60 days ago

Verified a81a5695

SOLVE_TRI extension to more dimensions (#17793)

pwilkin committed 60 days ago

Verified 53ecd4fd

ggml-alloc : fix reuse-parent logic for misaligned sizes (#17884)

ggerganov committed 60 days ago

Verified c6f6e4f9

batch : fix sequence id ownership (#17915)

ggerganov committed 60 days ago

Verified d9f8f606

docs: use port 8080 in Docker examples (#17903)

utsumi-fj committed 60 days ago

Verified e4ae3833

ggml-hexagon: fix `rope` failure at `test-backend-ops` (#17565)

chraac committed 61 days ago

Verified 34ce48d9

ci: fix riscv64-native build (#17916)

CISC committed 61 days ago

Verified 45e350e3

mtmd: some small clean up (#17909)

ngxson committed 61 days ago

Verified c6b2c931

cli: enable jinja by default (#17911)

ngxson committed 61 days ago

Verified 34a6d869

server: add presets (config) when using multiple models (#17859)

ServeurpersoCom committed 61 days ago

Verified f32ca51b

Fix race conditions in threadpool when dealing with dynamic/frequent n_threads changes (#17748)

max-krasnyansky committed 61 days ago

Verified e1f49219

ggml : remove GGML_KQ_MASK_PAD constant (#17910)

ggerganov committed 61 days ago

Verified 4dff236a

cuda : add missing support check for xielu (#17895)

CISC committed 61 days ago

Verified 4df6e859

cli: new CLI experience (#17824)

ngxson committed 61 days ago

Verified 6c213177

model : Qwen3-Next-80B-A3B has 48 layers (#17898)

EZForever committed 61 days ago

Verified b6777218

docs : update opencl ops (#17904)

lhez committed 61 days ago

Verified 2d2e1030

CUDA: fix unpadded strides in MMA FA kernel (#17891)

JohannesGaessler committed 61 days ago

Verified 17f7f4ba

convert: allow using quantized Mistral weight (#17889)

ngxson committed 61 days ago

Verified 9e79b011

fix softmax for iGPU (#17838)

NeoZhangJianyu committed 61 days ago

Verified 2e9eab80

common : add parser for ministral/mistral large 3/devstral 2 (#17713)

aldehir committed 62 days ago

Verified 2fbe3b7b

docs : update cpu and cuda ops (#17890)

CISC committed 62 days ago

Verified 63391852

metal: SSM kernel improvements (#17876)

gabe-l-hart committed 62 days ago

Verified 086a63e3

Add DIAG for CUDA (#17873)

pwilkin committed 62 days ago

Verified b6350926

docs: clarify that CPU support should be first (#17886)

JohannesGaessler committed 62 days ago

Verified 48f47565

ggml : Provide macos-specific backtrace printing to avoid terminal death (#17869)

gabe-l-hart committed 62 days ago

Verified 02e409a5

metal : print node names for debugging (#17882)

ggerganov committed 62 days ago

Verified 6b82eb78

ggml : allow fill node alloc inplace (#17870)

CISC committed 62 days ago

Verified 86a3f0fa

cmake: fix Mach-O current version number (#17877)

Rhys-T committed 62 days ago

Verified 63908b63

model : nit, DeepSeek V1 MoE is 16B and GigaChat is 20B (#12652)

CISC committed 62 days ago

Verified 42b12b56

Older