ggml-org/llama.cpp

Pull Requests Commits
ggml-cuda: fix unused parameter in non-HIP builds
Victor-Loos committed 7 days ago
624617ac
ggml-cuda: avoid ROCm_Host compute on HIP integrated GPUs
Victor-Loos committed 7 days ago
52ff274b
Add tensor name to JSON output
pwilkin committed 7 days ago
5efd7400
tentative Metal support
pwilkin committed 7 days ago
4a4d3510
Add missing unrolls
pwilkin committed 7 days ago
7bbb7a17
Fix braces
pwilkin committed 7 days ago
bd6ec8e9
Fix FATTN profiling
pwilkin committed 7 days ago
c6dec166
Converge implementation with export-graph-ops
pwilkin committed 7 days ago
f6919937
Add missing op parameters to the profiler; add support for test-backend-ops to run performance tests with exactly the tensor shapes from the run
pwilkin committed 7 days ago
81f8b35d
docs, pass copy details
pwilkin committed 7 days ago
085847cd
fix mul_mat_id stats, add throughput stat, add envvar trigger, add concurrent mode fix
pwilkin committed 7 days ago
b6e0078b
fix builds, integrate vulkan profiler, fix copy events, fix export
pwilkin committed 7 days ago
1608f866
Fix more missing backend stuff (and Python errors)
pwilkin committed 7 days ago
b00e3a53
add second dimension to reported tensors, fix Mac build, add missing initializer to all backends
pwilkin committed 7 days ago
834067f6
feat: cool profiler thingy
pwilkin committed 7 days ago
19028b55
ci : add test-llama-archs tensor split for Metal (#27598)
ggerganov committed 7 days ago
Verified 95b8e33e
contrib : recommend waiting for CI before merging (#27603)
nikwen committed 7 days ago
Verified a278dcef
server : add LLAMA_SERVER_SLOTS_N_DIFF (#27600)
ggerganov committed 7 days ago
Verified e8eed452
common : skip device_info loop if it's not going to be printed (#26692)
wolfpld committed 7 days ago
Verified ba8e0edd
DeepseekV4: fix rollback with multi-seq (#26756)
am17an committed 7 days ago
Verified b0539c43
[Tensor parallel] Fix meta tensor split state propagation (#27574)
gaugarg-nv committed 7 days ago
Verified d3371929
ui: Chat Conversation Tabbed navigation (#27263)
allozaur committed 7 days ago
Verified 8144f319
vendor : update subprocess.h (#27409)
cabelo committed 7 days ago
Verified 6657ded4
cuda : add POOL_1D support (#27573)
amankarki151 committed 7 days ago
Verified 29ea9412
common: json.h: fix clang lto (#27575)
ngxson committed 8 days ago
Verified 70adb1b4
vulkan : added the PAD_REFLECT_1D operation (#26586)
safiullah3915 committed 8 days ago
Verified 3f545bec
mtmd: use ggml_rope_set_offset (#27521)
ngxson committed 8 days ago
Verified b21e4de7
common: add json.h abstraction (#27511)
ngxson committed 8 days ago
Verified d9f918d2
fit: also take into account n_streams (#27496)
ngxson committed 8 days ago
Verified 2fb989b9
arg: remove -no-cnv from cli [no ci] (#27542)
ngxson committed 8 days ago
Verified 9fee29e9
Terms Of Use Privacy Policy Refund Policy Impressum