ochafik/llama.cpp

Pull Requests Commits

fix: use resolved URL directly in download

ochafik committed 180 days ago

fa874b06

download: add file size queries and disk space projections

ochafik committed 180 days ago

cbad19e2

tools: add llama-download for model management

ochafik committed 180 days ago

132460bf

ggml-cuda: fix regex for arch list (#18371)

am17an committed 180 days ago

Verified 85c40c9b

cuda: optimize cumsum cub path (#18362)

am17an committed 180 days ago

Verified 83b3b1c2

ggml-cuda: fix blackwell native builds (#18361)

am17an committed 180 days ago

Verified b0fb0f0a

CANN: Add support for CONV_TRANSPOSE_1D when kernel size > 255 (#17934)

Intellouis committed 180 days ago

Verified e68c19b0

ggml : optimize cuda cumsum fallback kernel (#18343)

Aadeshveer committed 181 days ago

Verified c54bba86

server: (router) add stop-timeout option (#18350)

ngxson committed 181 days ago

Verified f5acfb2f

model: support MiMo-V2-Flash (#18328)

ngxson committed 181 days ago

Verified 4cbafad4

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

Aadeshveer committed 181 days ago

Verified c1842842

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

am17an committed 181 days ago

Verified c8a2417d

model : support for LlamaBidirectionalModel architecture (#18220)

sfallah committed 181 days ago

Verified 54132f1b

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

jeffbolznv committed 181 days ago

Verified 2a9ea202

CANN : refactor ACL graph cache (#17752)

wangweixuan committed 181 days ago

Verified ce7a6dc0

docs: Fix typos in SYCL documentation (#18269)

yoka committed 181 days ago

Verified 1ce0126b

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

0cc4m committed 181 days ago

Verified 7f459c98

CANN: Uses yarn_ramp cache in ROPE (#17725)

TianHao324 committed 182 days ago

Verified cf2ffc02

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

ddh0 committed 182 days ago

Verified 10355dc7

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

ngxson committed 182 days ago

Verified 5ee4e43f

webui: apply webui_settings on first load (#18223)

ServeurpersoCom committed 182 days ago

Verified 5b6c9bc0

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

ngxson committed 182 days ago

Verified 849d0211

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

danbev committed 182 days ago

Verified 8e3ead6e

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

struct committed 182 days ago

Verified 12ee1763

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

chraac committed 183 days ago

Verified ed759777

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

danbev committed 183 days ago

Verified 847c35f7

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

arthw committed 183 days ago

Verified a6a552e4

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

Alessandro98-git committed 183 days ago

Verified 96e33a81

model : Granite Embedding support (#15641)

ryan-mangeno committed 183 days ago

Verified dfc959b8

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

compilade committed 183 days ago

Verified 8f488073

Older