Skip to content

Actions: ngxson/llama.cpp

Actions

Release

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
release.yml
1,060 workflow runs
1,060 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

cuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs (#14741) Release #360: Commit 021cc28 pushed by ngxson
27m 45s master
master
27m 45s
graph : avoid huge warm-up graphs for MoE models (#14753) Release #359: Commit d498af3 pushed by ngxson
25m 1s master
master
25m 1s
model : fix build after merge conflict (#14754) Release #358: Commit eacdeb5 pushed by ngxson
21m 50s master
master
21m 50s
model : add EXAONE 4.0 support (#14630) Release #357: Commit e0cb5c5 pushed by ngxson
12m 43s master
master
12m 43s
CUDA: set_rows + cpy.cu refactor (#14712) Release #356: Commit f9a31ee pushed by ngxson
21m 50s master
master
21m 50s
graph : refactor context to not pass gf explicitly (#14629) Release #355: Commit 8f974bc pushed by ngxson
14m 20s master
master
14m 20s
graph : Pass the graph placeholder message in debug mode (#14748) Release #354: Commit 09651d0 pushed by ngxson
17m 34s master
master
17m 34s
use max work group size for device to replace the magic number (#14732) Release #353: Commit 349ea79 pushed by ngxson
21m 51s master
master
21m 51s
model: add Ernie 4.5 MoE support (#14658) Release #352: Commit cb887f1 pushed by ngxson
23m 16s master
master
23m 16s
kv-cache : fix k-shift for multiple streams (#14742) Release #351: Commit d6fb3f6 pushed by ngxson
13m 44s master
master
13m 44s
llama : reuse compute graphs (#14482) Release #350: Commit 01612b7 pushed by ngxson
18m 5s master
master
18m 5s
llama : fix parallel processing for lfm2 (#14705) Release #349: Commit 086cf81 pushed by ngxson
20m 18s master
master
20m 18s
kv-cache : opt mask set input (#14600) Release #348: Commit d9b6910 pushed by ngxson
47m 50s master
master
47m 50s
batch : fix uninitialized has_cpl flag (#14733) Release #347: Commit ad57d3e pushed by ngxson
25m 36s master
master
25m 36s
llama : fix parameter order for hybrid memory initialization (#14725) Release #346: Commit 496957e pushed by ngxson
22m 5s master
master
22m 5s
ggml: Add initial WebGPU backend (#14521) Release #345: Commit 21c0217 pushed by ngxson
25m 4s master
master
25m 4s
model : support output bias for qwen2 (#14711) Release #344: Commit b0f0ecc pushed by ngxson
23m 12s master
master
23m 12s
llama : add high-throughput mode (#14363) Release #343: Commit 225e7a1 pushed by ngxson
28m 19s master
master
28m 19s
Support diffusion models: Add Dream 7B (#14644) Release #342: Commit ab14019 pushed by ngxson
19m 19s master
master
19m 19s
ggml : add asserts (#14720) Release #341: Commit 6497834 pushed by ngxson
32m 25s master
master
32m 25s
server : pre-calculate EOG logit biases (#14721) Release #340: Commit 6ffd4e9 pushed by ngxson
17m 29s master
master
17m 29s
llama : fix parallel processing for plamo2 (#14716) Release #339: Commit e4841d2 pushed by ngxson
23m 7s master
master
23m 7s
server : fix handling of the ignore_eos flag (#14710) Release #338: Commit 538cc77 pushed by ngxson
18m 35s master
master
18m 35s
llama: add LLAMA_API to deprecated llama_kv_self_seq_div (#14708) Release #337: Commit 79e0b68 pushed by ngxson
18m 42s master
master
18m 42s
model : add Kimi-K2 support (#14654) Release #336: Commit 4a4f426 pushed by ngxson
27m 11s master
master
27m 11s
ProTip! You can narrow down the results and go further in time using created:<2025-07-15 or the other filters available.