ggerganov/whisper.cpp

Pull Requests Commits

docs : fix duplicate word typo in VAD section (#3670)

cluster2600 committed 4 days ago

Verified 21411d81

talk-llama : sync llama.cpp

ggerganov committed 8 days ago

364c77f4

ggerganov committed 8 days ago

83f2ed19

models : optimize qwen3next graph (llama/19375)

ggerganov committed 8 days ago

4ac70ce7

ggml : fix GGML_DEBUG with OpenMP (llama/19599)

angt committed 8 days ago

226e8c04

metal : fix ACC op (llama/19427)

ggerganov committed 8 days ago

fbdac511

vulkan: support L2_NORM with contiguous rows (llama/19604)

jeffbolznv committed 8 days ago

cc448def

vulkan: support GGML_OP_SET (llama/19584)

jeffbolznv committed 8 days ago

197e9ab6

vulkan: Add vendor id for Qualcomm drivers (llama/19569)

strongtz committed 8 days ago

fc6bbab8

hexagon: further optimizations and refactoring for flash attention (llama/19583)

max-krasnyansky committed 8 days ago

e6476d4c

vulkan: restore -inf check in FA shaders (llama/19582)

jeffbolznv committed 8 days ago

ec57bf40

Fix wrong memcpy length for block_interleave == 4 (llama/19575)

Alcpz committed 8 days ago

e8a25654

fix vulkan ggml_acc only works in 3d but not 4d (llama/19426)

ymcki committed 8 days ago

628b545b

CUDA: loop over ne2*ne3 in case it overflows (llama/19538)

am17an committed 8 days ago

58e3d5a4

CUDA: Do not mutate cgraph for fused ADDs (llama/19566)

ORippler committed 8 days ago

3eb4905a

metal : improve concurrency (llama/19555)

ggerganov committed 8 days ago

0e94faa1

metal : support GGML_OP_SET (llama/19548)

ggerganov committed 8 days ago

c5325e50

hexagon: fix typo in vtcm_needs_release (llama/19545)

FanShupei committed 8 days ago

195af60a

opencl: add basic support for q4_1 (llama/19534)

lhez committed 8 days ago

9f87eecc

metal : update sum_rows kernel to support float4 (llama/19524)

ggerganov committed 8 days ago

d8e3e2ef

Add a workaround for compilation with ROCWMMA_FATTN and gfx9 (llama/19461)

superm1 committed 8 days ago

39b5f414

hexagon: further optimization and tuning of matmul and dot kernels (llama/19407)

max-krasnyansky committed 8 days ago

30420567

opencl: add general Q6_K mm and Q4_K mv (llama/19347)

lhez committed 8 days ago

0326fd37

ggml : unary ops support non-cont src0 + metal F16 unary ops (llama/19511)

ggerganov committed 8 days ago

f3e78985

metal : extend l2_norm support for non-cont src0 (llama/19502)

ggerganov committed 8 days ago

3ffa1fd8

hexagon: Add ARGSORT, DIV, SQR, SQRT, SUM_ROWS, GEGLU (llama/19406)

max-krasnyansky committed 8 days ago

09587ceb

ggml : extend bin bcast for permuted src1 (llama/19484)

ggerganov committed 8 days ago

35043580

metal : consolidate unary ops (llama/19490)

ggerganov committed 8 days ago

de949fb1

CUDA : Update CCCL-tag for 3.2 to final release from RC (llama/19486)

ORippler committed 8 days ago

57c620b4

Plug memory leaks and free resources on shutdown (llama/19315)

nikhilJain17 committed 8 days ago

562255fd

Older