Skip to content

Actions: ngxson/llama.cpp

Actions

Server

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
server.yml
1,109 workflow run results
1,109 workflow run results

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

CUDA: faster Deepseek FA, add Turing support (#13435) Server #1520: Commit 6da34fa pushed by ngxson
13m 12s master
master
13m 12s
fix: Move build_inp_pos to the top of the graph section for build_gra… Server #1519: Commit 5e7d95e pushed by ngxson
13m 52s master
master
13m 52s
server : passthrough the /models endpoint during loading (#13535) Server #1518: Commit 0531744 pushed by ngxson
9m 27s master
master
9m 27s
server : fix cache_tokens bug with no cache_prompt (#13533) Server #1517: Commit 360a9c9 pushed by ngxson
14m 31s master
master
14m 31s
cmake: simplify vulkan shader test logic (#13263) Server #1516: Commit 09d13d9 pushed by ngxson
9m 0s master
master
9m 0s
vulkan: KHR_coopmat flash attention (#13506) Server #1515: Commit 24e86ca pushed by ngxson
9m 27s master
master
9m 27s
webui : use fflate for more deterministic gzip compress (#13525) Server #1514: Commit bb1681f pushed by ngxson
9m 49s master
master
9m 49s
webui: Allow pasting file from clipboard (#13526) Server #1513: Commit d486dd3 pushed by ngxson
9m 4s master
master
9m 4s
quantize : improve tensor-type pattern matching (#13033) Server #1512: Commit e5c834f pushed by ngxson
8m 56s master
master
8m 56s
clip : clip.h become private API (⚠️ breaking change) (#13510) Server #1511: Commit 71bdbdb pushed by ngxson
9m 53s master
master
9m 53s
metal : use FA-vec kernel up to batch size 20 (#13496) Server #1510: Commit f0995d2 pushed by ngxson
11m 49s master
master
11m 49s
mtmd : remove libllava, remove clip-quantize-cli (⚠️ breaking change)… Server #1509: Commit b472634 pushed by ngxson
13m 0s master
master
13m 0s
model : Granite MoE shared (#13269) Server #1508: Commit d590cd4 pushed by ngxson
10m 15s master
master
10m 15s
llama-bench : add defrag-thold, check for invalid ranges (#13487) Server #1507: Commit cf0a43b pushed by ngxson
10m 23s master
master
10m 23s
opencl: remove unnecessary assert for add (#13257) Server #1506: Commit f0d46ef pushed by ngxson
8m 44s master
master
8m 44s
clip : cap max image size 1024 for qwen vl model (#13478) Server #1505: Commit de4c07f pushed by ngxson
17m 21s master
master
17m 21s
llama/ggml: add LLM training support (#10544) Server #1504: Commit 10d2af0 pushed by ngxson
11m 29s master
master
11m 29s
context : fix state io for memory-less contexts (#13470) Server #1503: Commit 064cc59 pushed by ngxson
10m 2s master
master
10m 2s
server : allow content to be null in oaicompat_completion_params_pars… Server #1502: Commit 91159ee pushed by ngxson
10m 4s master
master
10m 4s
llama-bench : accept ranges for integer parameters (#13410) Server #1501: Commit 22cdab3 pushed by ngxson
8m 53s master
master
8m 53s
ggml : add mrope kernel for metal (#13457) Server #1500: Commit df84919 pushed by ngxson
8m 50s master
master
8m 50s
enable dpcpp nightly builds with libraries (#13406) Server #1499: Commit 1449214 pushed by ngxson
10m 9s master
master
10m 9s
mtmd : Use RMS norm for InternVL 3 38B and 78B mmproj (#13459) Server #1498: Commit c104023 pushed by ngxson
8m 50s master
master
8m 50s
tools : fix uninitialized llama_batch in server (#13436) Server #1497: Commit 9a390c4 pushed by ngxson
10m 26s master
master
10m 26s
CUDA: fix crash with partial offloading of MoE (#13439) Server #1496: Commit 7474e00 pushed by ngxson
9m 16s master
master
9m 16s