ggml-org/whisper.cpp

Pull Requests Commits

talk-llama : sync llama.cpp

ggerganov committed 67 days ago

Verified fcaa8f8a

ggerganov committed 67 days ago

Verified 82ee376b

llama.android : Rewrite Android binding (w/o cpu_features dep) (llama/17413)

naco-siren committed 67 days ago

Verified 7e9b6ebb

ggml : use WARP_SIZE/2 for argmax reduction offset (llama/18092)

Aadeshveer committed 67 days ago

Verified 6829ed71

ggml-hexagon: mm for mtmd (llama/17894)

joeldushouyu committed 67 days ago

Verified 5cac2247

metal: use shared buffers on eGPU (llama/17866)

jdemeule committed 67 days ago

Verified 90e623c2

llama: automatically set parameters not set by the user in such a way that maximizes GPU utilization (llama/16653)

JohannesGaessler committed 67 days ago

Verified 542f01ec

Support gpt-oss by OPs add-id, mul_mat for mxfp4, swiglu_oai (llama/17826)

NeoZhangJianyu committed 67 days ago

Verified 6f0e8fec

vulkan: fix mul_mat_vec_iq1_s formatting (llama/18026)

0cc4m committed 67 days ago

Verified 631c3b39

vulkan: Fix data race/hang in scalar/cm1 flash attention (llama/17887)

jeffbolznv committed 67 days ago

Verified 72c24f6e

vulkan: improve mul_mat_vec_iq1_s speed (llama/17874)

lovedheart committed 67 days ago

Verified a2380a38

vulkan: faster q6_k matmul (llama/17813)

netrunnereve committed 67 days ago

Verified 9a211ac9

ggml : arm repack fix build (llama/0)

ggerganov committed 67 days ago

Verified e500fa6c

vulkan: support get_rows for i32 (llama/17941)

jeffbolznv committed 67 days ago

Verified c62c6104

vulkan: support GGML_OP_DIAG (llama/17893)

jeffbolznv committed 67 days ago

Verified 5cb35693

vulkan: Multi-pass softmax for large number of cols (llama/17892)

jeffbolznv committed 67 days ago

Verified 96fa6388

vulkan: Allow non-pow2 n_experts in topk_moe (llama/17872)

jeffbolznv committed 67 days ago

Verified 4de56992

CUDA: fix overflow in MMA kernel without stream-k (llama/17939)

JohannesGaessler committed 67 days ago

Verified 8d05cf47

cann : fix ops broken by circular padding guard (llama/17825)

CISC committed 67 days ago

Verified 240cb2db

ggml-cpu : fix RISC-V Q4_0 repack select and RVV feature reporting (llama/17951)

ixgbe committed 67 days ago

Verified a21115d3

HIP: enable mmf for RDNA3 (llama/17879)

zhang-hui-yulo committed 67 days ago

Verified 951f8a97

SOLVE_TRI extension to more dimensions (llama/17793)

pwilkin committed 67 days ago

Verified 43be89ef

build: link whisper target against Threads::Threads for FreeBSD support (#3568)

rsmirnov90 committed 67 days ago

Verified 3e79e73e

server: allow custom temp directory for ffmpeg (#3564)

socram8888 committed 71 days ago

Verified 2551e4ce

ggml : arm repack fix build (#0)

ggerganov committed 71 days ago

Verified f0c9017a

talk-llama : sync llama.cpp

ggerganov committed 71 days ago

Verified 179d8b1c

ggerganov committed 71 days ago

Verified 48cdc06e

whisper : adjust to ggml changes (#0)

ggerganov committed 71 days ago

Verified 72714d16

cmake : set `CMAKE_RUNTIME_OUTPUT_DIRECTORY` for non standalone build (ggml/1394)

HerrCai0907 committed 71 days ago

Verified 324dd21d

ggml-alloc : fix reuse-parent logic for misaligned sizes (llama/17884)

ggerganov committed 71 days ago

Verified 1da1a686

Older