ngxson/llama.cpp

Pull Requests Commits

unload_lru: also wait for unload to complete

ngxson committed 17 hours ago

0c2dc02a

ngxson committed 19 hours ago

9471e76a

also allow stop while loading

ngxson committed 19 hours ago

7e61df6e

server: (router) add stop-timeout option

ngxson committed 19 hours ago

88d6e710

fit-params : fix race condition in fit-params output (#18276)

Aadeshveer committed 20 hours ago

Verified c1842842

CUDA: experimental native mxfp4 support for blackwell (#17906)

am17an committed 21 hours ago

Verified c8a2417d

model : support for LlamaBidirectionalModel architecture (#18220)

sfallah committed 22 hours ago

Verified 54132f1b

vulkan: fix command buffer corruption in ggml_backend_vk_event_wait (#18302)

jeffbolznv committed 1 day ago

Verified 2a9ea202

CANN : refactor ACL graph cache (#17752)

wangweixuan committed 1 day ago

Verified ce7a6dc0

docs: Fix typos in SYCL documentation (#18269)

yoka committed 1 day ago

Verified 1ce0126b

vulkan: use fewer FA rows for small cache runs (#18280)

0cc4m committed 1 day ago

Verified 7f459c98

CANN: Uses yarn_ramp cache in ROPE (#17725)

TianHao324 committed 1 day ago

Verified cf2ffc02

common: add `LLAMA_ARG_OVERRIDE_TENSOR` env var for `-ot` arg (#18267)

ddh0 committed 1 day ago

Verified 10355dc7

server: return_progress to also report 0% processing state (#18305)

ngxson committed 1 day ago

Verified 5ee4e43f

webui: apply webui_settings on first load (#18223)

ServeurpersoCom committed 1 day ago

Verified 5b6c9bc0

server: fix crash with model not having BOS/EOS (#18321)

ngxson committed 1 day ago

Verified 849d0211

model-conversion : add device option to run-org-model.py (#18318)

danbev committed 1 day ago

Verified 8e3ead6e

rpc : add check for rpc buffer type (#18242)

struct committed 2 days ago

Verified 12ee1763

ggml-hexagon: create generalized functions for cpu side op (#17500)

chraac committed 2 days ago

Verified ed759777

model-conversion : add trust_remote_code for embedding scripts (#18288)

danbev committed 2 days ago

Verified 847c35f7

[SYCL] replace llama-cli by llama-completion to rm the impact to test script (#18290)

arthw committed 2 days ago

Verified a6a552e4

model : fix div-by-zero for Nemotron V2 (#18309)

Alessandro98-git committed 2 days ago

Verified 96e33a81

model : Granite Embedding support (#15641)

ryan-mangeno committed 2 days ago

Verified dfc959b8

gguf-py : do not align the data start offset (#18291)

compilade committed 2 days ago

Verified 8f488073

ggml-hexagon: gelu optimization (#18151)

joeldushouyu committed 2 days ago

Verified bf6bc3c1

gen-docs: automatically update markdown file (#18294)

ngxson committed 2 days ago

Verified 179fd82a

llamafile: add rvv support for sgemm kernels (#18199)

taimur-10x committed 2 days ago

Verified d34d5ca1

opencl: unpack q4_0 for adreno in get_tensor (#18278)

lhez committed 2 days ago

Verified eb492bf4

vulkan: Extend rope fusions to allow mrope (#18264)

jeffbolznv committed 2 days ago

Verified e3b35ddf

server: prevent data race from HTTP threads (#18263)

ngxson committed 2 days ago

Verified 6ce863c8

Older