ggerganov/llama.cpp

Pull Requests Commits

Revert "tests: serialize the router tests that download the same model"

ServeurpersoCom committed 1 day ago

8d79143a

tests: serialize the router tests that download the same model

ServeurpersoCom committed 1 day ago

c368a4a9

Revert "download: keep concurrent downloads of one blob apart"

ServeurpersoCom committed 1 day ago

6f4d5bfe

download: keep concurrent downloads of one blob apart

ServeurpersoCom committed 1 day ago

917b83f1

fix Windows build

ServeurpersoCom committed 1 day ago

1eb98003

server: refactor subproc handling

ngxson committed 1 day ago

5ceba566

cmake : skip PCH for llama-server PCH when using MSVC (#28763)

danbev committed 1 day ago

Verified 8ea29024

ci : run test-backend-ops as a dedicated ci/run.sh test (#28740)

ggerganov committed 1 day ago

Verified b78a39a2

tests: extend test-quantize-fns to test nrc=2 (i8mm) kernels (#16234)

Rohanjames1997 committed 1 day ago

Verified 982937a3

tests: tolerate a shared pool abort in test_completion_unified (#28759)

ServeurpersoCom committed 1 day ago

Verified 8172e657

ggml: skip 0-sized ids tensor when offloading selected experts (#28739)

am17an committed 1 day ago

Verified 43f3dda6

metal : skip the empty half of the mul_mm_id token tile (#28301)

masterFoad committed 2 days ago

Verified 5bda51bf

cmake : add PCH and unity build to improve build times (#28091)

danbev committed 2 days ago

Verified 3bcfeb70

common : fix typo in speculative.cpp comment [no ci] (#28750)

danbev committed 2 days ago

Verified 1dfe94e0

metal : single-source fusion table + fusion debug rework (#28164)

ggerganov committed 2 days ago

Verified a2878d30

metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 (#28692)

masterFoad committed 2 days ago

Verified aac81023

model : fix MTP context kv cache allocation for deepseek2, glm4moe, … (#28630)

LoganChu committed 2 days ago

Verified 5cdd3d1d

server: fix speculation after an image (#28715)

jesdga95 committed 2 days ago

Verified b0dcb819

CUDA/HIP: Flash Attention tuning (gfx1201) (#28102)

pwilkin committed 2 days ago

Verified 16378d93

ci : key cache to sanitizer matrix (#28708)

CISC committed 2 days ago

Verified 451b89ba

vulkan: fix data race and OOB access in argsort(large) (#28705)

jeffbolznv committed 2 days ago

Verified 481c65f0

opencl: add A8 Q4_0 mm binary kernel support (#28268)

shaofeiqi committed 2 days ago

Verified df03399b

vulkan: use CPU writes in ggml_backend_vk_cpy_tensor_async if the context is idle (#28618)

jeffbolznv committed 2 days ago

Verified 28ff0958

vulkan: use add_alloc_dep to enable topk_moe fusion for prefill (#28422)

jeffbolznv committed 2 days ago

Verified 50182a53

vulkan: small M matrix optimizations for qwen (#28457)

jeffbolznv committed 2 days ago

Verified 6788edb4

ci : add self-hosted-gpu-cuda and server-sanitize to hf-jobs (#28693)

CISC committed 2 days ago

Verified 52d42686

ci : Update WoA CUDA 13.4 release to use 13.4.1 GA redistributables (#28687)

shivamkumard-ctrl committed 2 days ago

Verified 18c17b4d

spec: fix failed to decode mtmd chunk with DFlash (#28587)

jesdga95 committed 2 days ago

Verified fa676981

scripts : use sed instead of grep for version parsing [no ci] (#28700)

danbev committed 2 days ago

Verified 41fc7584

models: clean up some dead switch branches in old models (#28669)

iggy committed 2 days ago

Verified d344123f

Older