ngxson/llama.cpp

Pull Requests Commits

Revert "DEMO repack"

ngxson committed 69 days ago

f46686b5

ngxson committed 69 days ago

caf0e423

fix device error

ngxson committed 69 days ago

489a7b8c

fix lazy loading

ngxson committed 69 days ago

ed7b7c72

rm redundant code

ngxson committed 69 days ago

ab0b550a

convert: add dequant function for compressed_tensor (kimi-k2-thinking)

ngxson committed 69 days ago

1bd57a3b

ggml-cpu : optimize RVV q2_k and q3_k kernels (#16887)

xctan committed 69 days ago

Verified 7f09a680

CUDA: fix crash on uneven context without FA (#16988)

JohannesGaessler committed 69 days ago

Verified aa374175

metal : initial Metal4 tensor API support (#16634)

ggerganov committed 69 days ago

Verified 5b180c3d

server : disable checkpoints with mtmd (#17045)

ggerganov committed 69 days ago

Verified b7f9010d

clip: implement minicpm-v sinusoidal embd using GGML (#17036)

ngxson committed 69 days ago

Verified 4882f0ff

sycl: add CONCAT operator support (#16047)

ye-NX committed 69 days ago

Verified 9d7c518d

docs: explain CUDA 11 compilation [no ci] (#16824)

JohannesGaessler committed 70 days ago

Verified 22c8c3c6

ggml-hexagon: graceful fallback for older socs where rpcmem_alloc2 and FASTRPC_GET_URI is unsupported (#16987)

l3utterfly committed 70 days ago

Verified 6db3d1ff

improve CUDA cpy memory bandwidth when copying transposed tensor (#16841)

bssrdf committed 70 days ago

Verified 230d1169

vulkan: Fix GGML_VULKAN_CHECK_RESULTS to better handle fusion (#16919)

jeffbolznv committed 70 days ago

Verified a44d7712

examples(gguf): GGUF example outputs (#17025)

gabe-l-hart committed 70 days ago

Verified 5886f4f5

mtmd: allow QwenVL to process larger image by default (#17020)

ngxson committed 70 days ago

Verified 92bb84f7

server : do not default to multiple slots with speculative decoding (#17017)

ggerganov committed 70 days ago

Verified 13b339bc

mtmd: improve struct initialization (#16981)

ngxson committed 70 days ago

Verified 2f0c2db4

docs: Clarify the endpoint that webui uses (#17001)

openingnow committed 70 days ago

Verified fd2f84f4

model : add openPangu-Embedded (#16941)

Lpzhan931 committed 71 days ago

Verified 9f052478

ggml webgpu: minor set rows optimization (#16810)

reeselevine committed 71 days ago

Verified 03ea0417

ggerganov committed 71 days ago

cdabeb2c

ggml : fix conv2d_dw SVE path (ggml/1380)

ggerganov committed 71 days ago

852ce518

CUDA: update ops.md (#17005)

mnehete32 committed 71 days ago

Verified 9aa63374

opencl: update doc (#17011)

lhez committed 71 days ago

Verified 5e90233b

refactor: replace sprintf with snprintf for safer string handling in dump functions (#16913)

chraac committed 71 days ago

Verified a5c07dcd

vulkan: remove the need for the dryrun (#16826)

jeffbolznv committed 71 days ago

Verified ad51c0a7

server : do context shift only while generating (#17000)

ggerganov committed 71 days ago

Verified 66d8eccd

Older