ggerganov/whisper.cpp

Pull Requests Commits

ruby : fix segmentation fault (#3591)

KitaitiMakoto committed 2 days ago

Verified 679bdb53

ggerganov committed 7 days ago

e9898ddf

ggml : bump version to 0.9.5 (ggml/1410)

ggerganov committed 7 days ago

ce03f8e7

talk-llama : sync llama.cpp

ggerganov committed 7 days ago

7359ac94

ggerganov committed 7 days ago

54fa8216

metal : add count_equal op (llama/18314)

gatbontonpc committed 7 days ago

8189f2cb

CUDA: fix KQ max calculation (llama/18487)

JohannesGaessler committed 7 days ago

2d250f80

metal : remove BF16 x F16 kernels (llama/18456)

ggerganov committed 7 days ago

5deaf8f2

sycl: add newline at the end of CMakeLists.txt (llama/18503)

am17an committed 7 days ago

46793319

Work around broken IntelSYCLConfig.cmake in Intel oneAPI 2025.x (llama/18345)

rrsathe committed 7 days ago

a3635494

kleidiai: add and integrate SVE 256-bit vector-length kernel (llama/18458)

chaxu01 committed 7 days ago

c9955367

CUDA: add log line when mxfp4 acceleration is used (llama/18483)

am17an committed 7 days ago

6d4aa96b

CUDA: fix replacment of bad archs in CMake (llama/18457)

JohannesGaessler committed 7 days ago

5765c5b0

CUDA: Blackwell features for non-native builds (llama/18436)

JohannesGaessler committed 7 days ago

d6cb2407

cuda: fix race condition in cumsum (llama/18448)

am17an committed 7 days ago

e49e88b2

HIP: Use mmq on MFMA devices for MUL_MAT_ID in cases where a lot of splits would be generated (llama/18202)

IMbackK committed 7 days ago

20f57299

Revert "ggml-cuda: use CMAKE_CUDA_ARCHITECTURES if set when GGML_NATIVE=ON (#18413)" (llama/18426)

am17an committed 7 days ago

b8d209f5

rpc: fix segfault on invalid endpoint format (llama/18387)

o7si committed 7 days ago

54fe9a64

cmake: Added more x86_64 CPU backends when building with `GGML_CPU_ALL_VARIANTS=On` (llama/18186)

bberberov committed 7 days ago

b3788ef7

ggml-cuda: use CMAKE_CUDA_ARCHITECTURES if set when GGML_NATIVE=ON (llama/18413)

QDelta committed 7 days ago

31fc2c37

opencl: allow resizing transpose buffers (llama/18384)

lhez committed 7 days ago

a800a3ac

ggml-cuda: Use same regex for GGML_NATIVE=OFF (llama/18407)

am17an committed 7 days ago

29f81554

vulkan: preprocess mul_mat_id experts and discard workgroups more quickly (llama/18352)

jeffbolznv committed 7 days ago

015b618d

vulkan: optimize decodeFuncB in coopmat2 mul_mat_id shader (llama/18349)

jeffbolznv committed 7 days ago

e37c8ed9

vulkan: Use BK=32 for coopmat2 mul_mat_id (llama/18332)

jeffbolznv committed 7 days ago

331c6ccd

vulkan: small dequantization improvements (llama/18380)

netrunnereve committed 7 days ago

35cb4abb

vulkan: Support UPSCALE w/antialias (llama/18327)

jeffbolznv committed 7 days ago

181e36f1

vulkan: handle rope with large number of rows (llama/18306)

jeffbolznv committed 7 days ago

67473fef

CANN: implement the SSM_CONV operator (llama/17737)

0Marble committed 7 days ago

33f75a88

ggml-cuda: fix regex for arch list (llama/18371)

am17an committed 7 days ago

51778354

Older