feat: support vad for addon.node (#3301)
f795870
unverified
Lin Xiaodong
linxiaodong
commited on
Spaces:
Sleeping
Sleeping
Commit History
sync : ggml
4966aed
talk-llama : sync llama.cpp
0ec1374
ggml : remove trailing whitespace (llama/0)
e37767f
opencl : add GEGLU, REGLU, SWIGLU (llama/14456)
d70ff9f
lhez
commited on
Add Conv2d for CPU (llama/14388)
68eb27a
metal : disable fast-math for some cpy kernels (llama/14460)
9d1185a
ggml-cpu: sycl: Re-enable exp f16 (llama/14462)
c0fcd7a
Romain Biessy
commited on
cmake : Remove redundant include path in CMakeLists.txt (llama/14452)
6b59b68
xiaobing318
commited on
scripts : make the shell scripts cross-platform (llama/14341)
9de52c8
Vedran Miletić
commited on
SYCL: disable faulty fp16 exp kernel (llama/14395)
b4969ff
Akarshan Biswas
commited on
ggml : fix unmerged GGML_FPxx_TO_FPxx refactoring (llama/14443)
f7995cb
Sigbjørn Skjæret
commited on
ggml : implement REGLU/GEGLU/SWIGLU ops (llama/14158)
add5c0f
vulkan: Add fusion support for RMS_NORM+MUL (llama/14366)
737f12d
CUDA: add bf16 and f32 support to cublas_mul_mat_batched (llama/14361)
c7936d3
vulkan: handle noncontig in the final case of ggml_vk_get_cpy_pipeline (llama/14378)
1c3b94c
vulkan: lock accesses of pinned_memory vector (llama/14333)
59dca4f
fix async_mode bug (llama/14432)
122b29a
vulkan: Fix GGML_VULKAN_SHADER_DEBUG_INFO (llama/14427)
a06c8ca
cmake: regen vulkan shaders when shaders-gen sources change (llama/14398)
7988638
bandoti
commited on
metal : add special-case mat-vec mul for ne00 == 4 (llama/14385)
724622d
metal : batch rows copy in a single threadgroup (llama/14384)
b4ff704
musa: enable fp16 mma (all) and cublas on qy2 (llama/13842)
e35329b
ggml-cpu: enable IBM NNPA Vector Intrinsics (llama/14317)
fea8f94
ggml : do not output unprintable characters on GGUF load failure (llama/14381)
e7b2e19
Sigbjørn Skjæret
commited on
sycl: GGML_SYCL_DISABLE_OPT on by default for all Intel Devices (llama/13973)
b25d3bf
Anton Mitkov
commited on
opencl: ref count `ggml_backend_opencl_context` and refactor profiling (llama/14254)
ae0c7b8
lhez
commited on
CUDA/HIP: optimize mmv paths taken for HIP devices (llama/14324)
1a9d2d3
CUDA: mul_mat_v support for batch sizes > 1 (llama/14262)
2d1e6e7
HIP: enable vec fattn on RDNA4 (llama/14323)
b6dc6a1
uvos
commited on
CUDA: add mean operation (llama/14313)
7cee55b
Add support for VK_EXT_debug_utils to add labels to Vulkan objects. (llama/13792)
2c3741a
Markus Tavenrath
commited on
metal : fix thread-safety (llama/14300)
2bd85b6
ggml-cpu : "align corners" for bilinear upscale/downscale (ggml/1285)
88e7829
ggml-quants : rename best_mad to best_error (ggml/1283)
cd9270f
ci : use selective copy for musa image (#3296)
e11fac2
unverified
ci: set fail-fast to false in docker.yml (#3294)
fc20738
unverified
ruby : add Whisper::VERSION (#3292)
6468892
unverified
whisper : add version function (#3289)
0b952d7
unverified
ci : add should_release variable (#3288)
7272681
unverified
docs : add cmake "-j" flag in README.md (#3284)
5fe3033
unverified
toboil-features
commited on
ci : add support for tag-based releases (#3287)
f21cf37
unverified
release : v1.7.6
5cade6e
unverified
bench : update benches
d4f72cd
unverified
bench : print system info before ctx check
835c3e8
unverified
stream : add nullptr check of whisper_context (#3283)
9f0c009
unverified
ci : enable main-cuda build (#3282)
9bee7f3
unverified
bindings.java : update java example (#3281)
f001158
unverified
Joas Dev
commited on
coreml : backport CoreML features to macos < 14 (#3255)
dc0917f
unverified
glaszig
commited on