Skip to content

Actions: ngxson/llama.cpp

Actions

Release

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
release.yml
1,033 workflow runs
1,033 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Vulkan: Add Integer Dot Product mul_mat_vec shader for legacy quants … Release #658: Commit 02c1813 pushed by ngxson
22m 45s master
master
22m 45s
ggml : WebGPU add TRANSPOSE and RESHAPE to supported ops (#15695) Release #657: Commit 77dee9d pushed by ngxson
15m 30s master
master
15m 30s
CUDA: fix build error from ambiguous __half conversions in conv2d (#1… Release #656: Commit b66df9d pushed by ngxson
32m 34s master
master
32m 34s
CANN: Optimize MUL_MAT_ID (#15658) Release #655: Commit b9382c3 pushed by ngxson
23m 51s master
master
23m 51s
server : enable /slots by default and make it secure (#15630) Release #654: Commit 0d161f0 pushed by ngxson
15m 17s master
master
15m 17s
llama : fix fattn reserve call n_seqs parameter (#15699) Release #653: Commit 2749662 pushed by ngxson
26m 30s master
master
26m 30s
llama : separate compute buffer reserve from fattn check (#15696) Release #652: Commit 9777032 pushed by ngxson
47m 29s master
master
47m 29s
vulkan: handle large sizes for get_rows (#15686) Release #651: Commit bbbf5ec pushed by ngxson
29m 11s master
master
29m 11s
vulkan: mul_mat_id coopmat2 optimizations (#15546) Release #650: Commit c37052a pushed by ngxson
33m 2s master
master
33m 2s
vulkan : remove unused portability_enumeration_ext variable (#15679) Release #649: Commit 5c16b9c pushed by ngxson
49m 4s master
master
49m 4s
vulkan: Allow fallback to sysmem memory when vidmem is full (#15649) Release #648: Commit b97c9ed pushed by ngxson
54m 31s master
master
54m 31s
vulkan: clamp matmul and FA results to the max finite value (#15652) Release #647: Commit 94e82c7 pushed by ngxson
36m 47s master
master
36m 47s
ggml: update kleidiai to v1.13.0 (#15663) Release #646: Commit 4d74393 pushed by ngxson
19m 7s master
master
19m 7s
llama: use FA + max. GPU layers by default (#15434) Release #645: Commit e81b8e4 pushed by ngxson
27m 53s master
master
27m 53s
CUDA: use FP32 arithmetic for conv2d (#15683) Release #644: Commit 38ad381 pushed by ngxson
26m 57s master
master
26m 57s
vulkan: Skip syncing for prealloc_y when it is reused (#15544) Release #643: Commit 696fccf pushed by ngxson
20m 23s master
master
20m 23s
CANN: FIx compiler warnings (#15661) Release #642: Commit ef47691 pushed by ngxson
18m 4s master
master
18m 4s
CUDA: fix bug in rms_norm fusion (#15660) Release #641: Commit 8101786 pushed by ngxson
22m 46s master
master
22m 46s
chat : Seed OSS thinking + tool call support (#15552) Release #640: Commit 60e5eee pushed by ngxson
20m 54s master
master
20m 54s
CUDA: fuse adds, fuse add with rms norm (#15631) Release #639: Commit 009b709 pushed by ngxson
17m 30s master
master
17m 30s
nvidia nemotron nano v2 (nemotronh) (#15507) Release #638: Commit e8d99dd pushed by ngxson
28m 44s master
master
28m 44s
fix: Compute the full sum in llama-eval-callback, not just the sum of… Release #637: Commit a8bca68 pushed by ngxson
27m 35s master
master
27m 35s
CUDA: add conv2d (#15635) Release #636: Commit c97dc09 pushed by ngxson
20m 11s master
master
20m 11s
ggml-cpu: fix invalid hsum build in debug s390x (#15634) Release #635: Commit 6c442f4 pushed by ngxson
36m 3s master
master
36m 3s
ggml : fix SSM_SCAN for n_groups > 1 (#15625) Release #634: Commit 7380414 pushed by ngxson
39m 15s master
master
39m 15s
ProTip! You can narrow down the results and go further in time using created:<2025-08-28 or the other filters available.