ggerganov/whisper.cpp

Pull Requests Commits

ggml : bump version to 0.9.4 (ggml/1363)

ggerganov committed 120 days ago

Verified 8a7f3b03

bench : update [no ci]

ggerganov committed 120 days ago

Verified e4bf87b0

ggerganov committed 120 days ago

Verified b57b9d3a

cuda : Enable CUDA Graph usage for Nemotron Nano v2 (NemotronH) (llama/16328)

anavp-nvidia committed 120 days ago

Verified 62b3b86e

metal : dynamic simdgroups for MV kernels (llama/16340)

ggerganov committed 120 days ago

Verified 78f85f2b

kleidiai : fix work size and threads sync for fp16 (llama/16246)

chaxu01 committed 120 days ago

Verified 01e86b69

ggml: riscv: add riscv spacemit backend (llama/15288)

alex-spacemit committed 120 days ago

Verified 35ebdf73

ggml-backend : add root cause in error message if loading backend library fails (llama/16172)

rlewczuk committed 120 days ago

Verified 94fe9bbe

bench : update [no ci] (#3439)

ggerganov committed 121 days ago

Verified 32be14f8

bench : warm-up all kernels (#3438)

ggerganov committed 121 days ago

Verified a77d11d9

ggml : remove oboslete files (#0)

ggerganov committed 121 days ago

Verified 22c12ee8

ci : add self-hosted workflows (#3437)

ggerganov committed 121 days ago

Verified d8cdcce8

whisper : remove ggml_mul_mat padding (#3436)

ggerganov committed 121 days ago

Verified b4909a6c

talk-llama : sync llama.cpp

ggerganov committed 121 days ago

Verified fcf0181e

ggerganov committed 121 days ago

Verified 404a9311

cmake : remove metal flag (llama/0)

ggerganov committed 121 days ago

Verified 32013827

ggml : check cuda and metal argsort limits and add test (llama/16323)

CISC committed 121 days ago

Verified 112e10f2

ggml : fix dependencies for ggml_set_rows (llama/16318)

ggerganov committed 121 days ago

Verified 7ce0a7bc

vulkan: Fix validation failure in quantized flash attention (llama/16292)

jeffbolznv committed 121 days ago

Verified a375e4c4

ggml : fix GGML_F32_VEC_FMA argument order in ggml_vec_mad1_f32 (llama/16307)

CISC committed 121 days ago

Verified 5c6e7956

vulkan: 64-bit im2col (llama/16135)

jeffbolznv committed 121 days ago

Verified 55d45edf

metal : extend mat-mat multiplication support (llama/16225)

ggerganov committed 121 days ago

Verified 0102733c

metal : fuse non-sequential nodes (llama/16102)

ggerganov committed 121 days ago

Verified 45976f28

vulkan: handle mat_mul with A matrix > 4GB (llama/16176)

jeffbolznv committed 121 days ago

Verified 91ab93b7

vulkan: support arbitrary KV dimension in flash attention (llama/16160)

jeffbolznv committed 121 days ago

Verified eb982dd7

vulkan : make the vulkan.hpp dynamic dispatcher instance private (llama/16224)

Acly committed 121 days ago

Verified bc1ac13c

CUDA: mul_mat_id for mmf for bs <= 64 for f16 and bs <= 32 for f32 (llama/16277)

am17an committed 121 days ago

Verified 85e4455c

CUDA: refactor and deduplicate vector FA kernels (llama/16208)

JohannesGaessler committed 121 days ago

Verified e856483c

vulkan: throw system error instead of SIGABRT during init on older devices (llama/16156)

DmyMi committed 121 days ago

Verified 88dd9e0d

vulkan: support GET_ROWS for k-quants (llama/16235)

jeffbolznv committed 121 days ago

Verified 97bd65f9

Older