ngxson/llama.cpp

Pull Requests Commits

Merge branch 'master' into xsn/server_tools

ngxson committed 19 days ago

8098f11f

ngxson committed 19 days ago

e4cc43a8

docs : rerun llama-gen-docs to include new CLI args (#20892)

EZForever committed 19 days ago

Verified 841bc203

add readme mention

ngxson committed 19 days ago

b648215e

Merge branch 'master' into xsn/server_tools

ngxson committed 19 days ago

7fbf8650

server: use httplib dynamic threads (#20817)

ngxson committed 19 days ago

Verified 31a5cf4c

ai : update gh permissions (#20895)

ggerganov committed 19 days ago

Verified e32d2438

webui: fix --webui-config-file settings not applied on load (#20823)

ServeurpersoCom committed 20 days ago

Verified c44a932c

metal: add CONV_3D (#19927)

Ra5hidIslam committed 20 days ago

Verified 177c7585

common/autoparser : detect reasoning markers when enable_thinking changes system prompt (#20859)

jhen0409 committed 20 days ago

Verified 7a0b6a63

CANN: add RoPE cache preload before ACL graph capture (#20747)

noemotiovon committed 20 days ago

Verified 07ff0005

fix(openvino): explicit memset in buffer_context allocation (#20857)

thedanhoffman committed 20 days ago

Verified cc18f965

opencl: add flattened Q4_K mv and general Q4_K mm (#20773)

shaofeiqi committed 20 days ago

Verified 84ffd0c1

mtmd: Add dynamic high-resolution image preprocessing for InternVL model (#20847)

bssrdf committed 20 days ago

Verified ec2b787e

mtmd : fix LightOnOCR image preprocessing (#20877)

DorianRudolph committed 20 days ago

Verified d3ac030a

server: allow router to report child instances sleep status (#20849)

ngxson committed 20 days ago

Verified 49bfddec

CUDA: fix BF16 FA compilation (#20865)

JohannesGaessler committed 20 days ago

Verified bd3f1d9d

jinja : refactor token advancement (#20864)

CISC committed 20 days ago

Verified 23c9182c

server: fix Host header (#20843)

kurnevsky committed 20 days ago

Verified 81bc4d3d

support bf16 and quantized type (#20803)

arthw committed 20 days ago

Verified f40a80b4

ggml-cuda: native bf16 flash attention for vec kernel (#20525)

eous committed 21 days ago

Verified db9d8aa4

[CUDA] Increase number of output elements per-thread block if the K-dimension is small (#20635)

gaugarg-nv committed 21 days ago

Verified ccb87fa3

misc : prefer ggml-org models in docs and examples (#20827)

ddh0 committed 21 days ago

Verified 3306dbae

common/grammar: fix grammar parsing issues to prevent stack overflow and hangs (#18604)

aagit committed 21 days ago

Verified 990e4d96

context : use n_embd_out for pooled embedding extraction (#20840)

extfs committed 21 days ago

Verified 212f4521

docs : explicit about banning accounts that violates policy (#19593)

ngxson committed 21 days ago

Verified 568aec82

fix(rpc): prevent division by zero in deserialize_tensor (#20712)

y198nt committed 21 days ago

Verified 2bcdddd5

Convert: Make NVFP4 and MXFP4 HF conversions say NVFP4/MXFP4 instead of BF16 (#20730)

michaelw9999 committed 21 days ago

Verified eac9c6ea

ci : switch from pyright to ty (#20826)

CISC committed 22 days ago

Verified 29b28a98

Add shader count for Intel Arc Pro B60 (#20818)

TheBlueMatt committed 22 days ago

Verified cea560f4

Older