ggerganov/whisper.cpp

Pull Requests Commits

CUDA: optimize FA for GQA + large batches (llama/12014)

JohannesGaessler committed 332 days ago

2d70cd36

cuda: Add Q5_1, Q5_0, Q4_1 and Q4_0 to F32 conversion support. (llama/12000)

gcp committed 332 days ago

98dab49b

CUDA: correct the lowest Maxwell supported by CUDA 12 (llama/11984)

PureJourney committed 332 days ago

b1385e9a

MUSA: support ARM64 and enable dp4a .etc (llama/11843)

BodhiHu committed 332 days ago

48f5e893

ggml-cpu: Add CPU backend support for KleidiAI library (llama/11390)

chaxu01 committed 332 days ago

dc21871f

ggml: aarch64: implement SVE kernels for q3_K_q8_K vector dot (llama/11917)

Vithulep committed 332 days ago

64a430bc

CUDA: use async data loading for FlashAttention (llama/11894)

JohannesGaessler committed 332 days ago

51a3580c

vulkan: implement several ops relevant for ggml_opt (llama/11769)

remyoudompheng committed 332 days ago

37a21dd4

vulkan: support multi/vision rope, and noncontiguous rope (llama/11902)

jeffbolznv committed 332 days ago

8a22a8b1

metal : fix the crash caused by the lack of residency set support on Intel Macs. (llama/11904)

halechan committed 332 days ago

fcbcad0c

metal : optimize dequant q6_K kernel (llama/11892)

akretz committed 332 days ago

4444db73

repo : update links to new url (llama/11886)

ggerganov committed 332 days ago

a7fc1038

vulkan: initial support for IQ1_S and IQ1_M quantizations (llama/11528)

remyoudompheng committed 332 days ago

1689aaf8

opencl: Fix rope and softmax (llama/11833)

lhez committed 332 days ago

4b48fe44

cuda : add ampere to the list of default architectures (llama/11870)

slaren committed 332 days ago

47cc043e

ggml: optimize some vec dot functions for LoongArch ASX (llama/11842)

MQ-mengqing committed 332 days ago

e3d9ffb9

vulkan: linux builds + small subgroup size fixes (llama/11767)

netrunnereve committed 332 days ago

e22d6983

llamafile: use member variable instead of constant for iq4nlt (llama/11780)

jmorganca committed 332 days ago

defe7312

musa: bump MUSA SDK version to rc3.1.1 (llama/11822)

yeahdongcn committed 332 days ago

4e07957b

ggml-cpu : add chunking support to mul_mat_id (llama/11666)

slaren committed 332 days ago

d2c5154b

ggml : x2 speed for WASM by optimizing SIMD (llama/11453)

ngxson committed 332 days ago

4fac43fe

HIP: Remove GCN from list of devices that avoid MMQ (llama/11831)

IMbackK committed 332 days ago

3be9670f

HIP: Switch to std::vector in rocblas version check (llama/11820)

IMbackK committed 332 days ago

86729fcd

cleanup: fix compile warnings associated with gnu_printf (llama/11811)

mtmcp committed 332 days ago

7fbca630

ggml : fix multi-threaded clamp_f32 (llama/11824)

Burton2000 committed 332 days ago

d597f83e

ggml-cpu: Fix duplicate MATMUL_INT8 (llama/11817)

ownia committed 332 days ago

e5edcc62

CUDA: fix CUDART_VERSION checks (llama/11821)

JohannesGaessler committed 332 days ago

556f773d

Fix #11802: Compile bug - RegQueryValueExA changed to RegQueryValueEx (llama/11803)

sheldonrobinson committed 332 days ago

91d02de3

CUDA: use arch list for compatibility check (llama/11775)

JohannesGaessler committed 332 days ago

1b67d72f

fix: typos in documentation files (llama/11791)

maximevtush committed 332 days ago

14d7c036

Newer Older