ggerganov/whisper.cpp

Pull Requests Commits

select utf8 codepage on windows

Judd committed 345 days ago

00ddb10f

fix building with MSVC + SDL2

Judd committed 345 days ago

b3a6018b

Fixes for Windows (#2790)

foldl committed 358 days ago

Verified d682e150

cmake : fix compile assumptions for power9/etc (#2777)

midnightmagic committed 359 days ago

Verified 46d07b9c

authors : update

ggerganov committed 360 days ago

Verified 33ea03f1

ggerganov committed 360 days ago

Verified dbcc669e

cmake: Add ability to pass in GGML_BUILD_NUMBER (ggml/1096)

ckastner committed 360 days ago

Verified 16245b35

readme : add maintenance roadmap

ggerganov committed 360 days ago

Verified 898c0cb9

ci : add stalebot

ggerganov committed 360 days ago

Verified eb9e5032

node : add max_len params in node addon (#2760)

billyct committed 360 days ago

Verified cadfc50e

talk-llama : sync llama.cpp

ggerganov committed 360 days ago

Verified 3f918323

coreml : always convert to "neuralnetwork" (#2770)

mgrachten committed 360 days ago

Verified cff8868b

ggerganov committed 360 days ago

90e3c5fc

ci : install git

ggerganov committed 360 days ago

e0f4cef8

ci : use ubuntu-22.04 instead of ubuntu-latest

ggerganov committed 360 days ago

23446098

cmake : sync cmake scripts

ggerganov committed 360 days ago

b8ab1263

ggerganov committed 360 days ago

edc5d926

scripts : fix sync paths

ggerganov committed 360 days ago

344b98a4

CUDA: fix Volta FlashAttention logic (llama/11615)

JohannesGaessler committed 360 days ago

dbeb7916

HIP: fix flash_attn_stream_k_fixup warning (llama/11604)

JohannesGaessler committed 360 days ago

fad28063

CUDA/HIP: add support for selectable warp size to mmv (llama/11519)

IMbackK committed 360 days ago

9906792e

HIP: add GGML_CUDA_CC_IS_* for amd familys as increasing cc archtectures for amd gpus are not supersets of eatch other (llama/11601)

IMbackK committed 360 days ago

c49ee07f

CUDA: use mma PTX instructions for FlashAttention (llama/11583)

JohannesGaessler committed 360 days ago

f8a83177

`ci`: use sccache on windows instead of ccache (llama/11545)

ochafik committed 360 days ago

85451e36

HIP: require at least HIP 5.5

IMbackK committed 360 days ago

43c744ce

HIP: Prepare reduction operators for wave 64

IMbackK committed 360 days ago

fc2e4449

CUDA/HIP: add warp_size to cuda_device_info

IMbackK committed 360 days ago

f41fdad2

vulkan: implement initial support for IQ2 and IQ3 quantizations (llama/11360)

remyoudompheng committed 360 days ago

80fa5762

vulkan: Catch pipeline creation failure and print an error message (llama/11436)

jeffbolznv committed 360 days ago

75e7d058

HIP: Supress transformation warning in softmax.cu

IMbackK committed 360 days ago

682a6f5f

Older