ggml-org/whisper.cpp

Pull Requests Commits

ggml-opencl : replace alloca() with std::vector (llama/29765)

angt committed 9 days ago

Verified c1c13908

cuda: guard the iq4_nl dequantize row kernel against short rows (llama/29683)

yeahdongcn committed 9 days ago

Verified bffb6e4b

Hexagon: optimize ALLREDUCE with support for safe scatter mode (llama/29757)

ebateni committed 9 days ago

Verified 6773ef87

ggml/gguf : fix integer overflow (llama/29384)

apach301 committed 9 days ago

Verified 02ea0c2e

ggml-et : remove useless alloca() (llama/29663)

angt committed 9 days ago

Verified 067bb06b

ggml : add BF16 unary, GLU, binary and scale ops (CPU, CUDA) (llama/29675)

am17an committed 9 days ago

Verified fd66b6ec

cpu: accept BF16 in src1 of mul_mat (llama/28937)

ServeurpersoCom committed 9 days ago

Verified c1b3fb10

openvino: serve GET_ROWS on a weight view from the base Constant (llama/28381)

ServeurpersoCom committed 9 days ago

Verified d0af734d

musa : define __CUDA_ARCH__ for device passes (llama/29508)

yeahdongcn committed 9 days ago

Verified 1cbf7e79

SYCL: reduce tensor allreduce sync with pinned host buffers (llama/29604)

Captain-Tripps committed 9 days ago

Verified d03bfd7f

ggml-zdnn: impl buffer reset, fix memory leaks (llama/29637)

taronaeo committed 9 days ago

Verified 66e8b957

Hexagon f16 activation ops (llama/29209)

cqderek committed 9 days ago

Verified b8051db8

gguf : reject tensor size that wraps after padding (llama/26979)

x14ngch3n committed 9 days ago

Verified 6e48a365

ggml : check row bounds in get_rows_back (llama/29575)

angt committed 9 days ago

Verified 14423995

hexagon: optimize concat op (llama/29673)

trivikram-reddy1 committed 9 days ago

Verified 8d2a2cb4

CUDA: bitonic argsort handles rows wider than one block (llama/28957)

ServeurpersoCom committed 9 days ago

Verified 45093fc8

ggml-cuda: HIP: optimize packed byte subtraction (`__vsubss4` -> `__vsub4`) (llama/29478)

thelittlefireman committed 9 days ago

Verified 482baacd

ci: add zdnn backend build but not test (llama/29541)

taronaeo committed 9 days ago

Verified 3b86cabd

opencl: fix get_tensor for q5_K adreno gemm_nonshuffle kernel (llama/29555)

lhez committed 9 days ago

Verified 9f12c9fe

vulkan: Tune GDN kernel, fix Intel performance (llama/29476)

0cc4m committed 9 days ago

Verified 5439f3d9

vulkan : Load F32 A matrix 2 at a time when its 2-aligned (llama/29254)

TheBlueMatt committed 9 days ago

Verified 6c07d821

vulkan: MOE aware mat_mul_id tile selection (llama/29182)

Ankk98 committed 9 days ago

Verified 8cc3ca48

ggml : fix c++ odr by properly using GGML_COMMON_DECL_CPP (llama/29504)

XZiar committed 9 days ago

Verified deae6f34

ggml : accumulate f16 dot products in f32 on AVX512-FP16 (llama/29545)

angt committed 9 days ago

Verified b97d4193

ggml : require input tensors to be GGML_OP_NONE (llama/29647)

ggerganov committed 9 days ago

Verified 4d4817ed

hexagon: add FP32 GELU_ERF and GEGLU_ERF support (llama/29631)

aparmp-quic committed 9 days ago

Verified 5e1914a7

ggml : collect all input tensors into graph_inputs (llama/29634)

ggerganov committed 9 days ago

Verified 763b67fd

ggml-zdnn: fix 0-row tensor crash (llama/29636)

taronaeo committed 9 days ago

Verified 84781290

ggml : speed up model loading (llama/29598)

angt committed 9 days ago

Verified f449de94

metal: FWHT perf optimizations (llama/29602)

bri-prism committed 9 days ago

Verified b30ef05f

Newer Older