ggerganov/llama.cpp

Pull Requests Commits

convert : use F32 for dequant of pack-quantized tensors

compilade committed 52 days ago

128118fd

convert : fix flake8 lint

compilade committed 52 days ago

3770d941

gguf-py : __pos__ is also unary

compilade committed 52 days ago

987862ad

convert : handle naive-quantized models

compilade committed 52 days ago

33dcb44a

convert : handle int-quantized models

compilade committed 52 days ago

d23bdd57

convert : handle compressed-tensors quant method

compilade committed 52 days ago

33dba6ce

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

xctan committed 52 days ago

Verified 7f09a680

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

JohannesGaessler committed 53 days ago

Verified aa374175

metal : initial Metal4 tensor API support (#16634)

ggerganov committed 53 days ago

Verified 5b180c3d

server : disable checkpoints with mtmd (#17045)

ggerganov committed 53 days ago

Verified b7f9010d

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

ngxson committed 53 days ago

Verified 4882f0ff

sycl: add CONCAT operator support (#16047)

ye-NX committed 53 days ago

Verified 9d7c518d

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

JohannesGaessler committed 53 days ago

Verified 22c8c3c6

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

l3utterfly committed 53 days ago

Verified 6db3d1ff

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

bssrdf committed 53 days ago

Verified 230d1169

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

jeffbolznv committed 53 days ago

Verified a44d7712

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

gabe-l-hart committed 53 days ago

Verified 5886f4f5

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

ngxson committed 54 days ago

Verified 92bb84f7

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

ggerganov committed 54 days ago

Verified 13b339bc

mtmd: improve struct initialization (#16981)

ngxson committed 54 days ago

Verified 2f0c2db4

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

openingnow committed 54 days ago

Verified fd2f84f4

model : add openPangu-Embedded (#16941)

Lpzhan931 committed 54 days ago

Verified 9f052478

ggml webgpu: minor set rows optimization (#16810)

reeselevine committed 54 days ago

Verified 03ea0417

ggerganov committed 54 days ago

cdabeb2c

ggml : fix conv2d_dw SVE path (ggml/1380)

ggerganov committed 54 days ago

852ce518

CUDA: update ops.md (#17005)

mnehete32 committed 54 days ago

Verified 9aa63374

opencl: update doc (#17011)

lhez committed 54 days ago

Verified 5e90233b

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

chraac committed 54 days ago

Verified a5c07dcd

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

jeffbolznv committed 54 days ago

Verified ad51c0a7

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

ggerganov committed 54 days ago

Verified 66d8eccd

Older