ggml-org/llama.cpp

Pull Requests Commits

ggml: add ops for WAN video model (cuda && cpu) (#15669)

leejet committed 6 days ago

Verified 0a1b3982

CANN: Fix precision issue on 310I DUO multi-devices (#15784)

hipudding committed 6 days ago

Verified 5421f63a

opencl: add hs=40 to FA (#15758)

rmatif committed 6 days ago

Verified 820bc985

CANN: fix acl_rstd allocation size in ggml_cann_rms_norm (#15760)

noemotiovon committed 6 days ago

Verified 239b60e8

vulkan: fix mmv subgroup16 selection (#15775)

0cc4m committed 6 days ago

Verified dff7551b

vulkan: don't use std::string in load_shaders, to improve compile time (#15724)

jeffbolznv committed 6 days ago

Verified 0fce7a12

vulkan : update ggml_vk_instance_validation_ext_available (#15666)

danbev committed 6 days ago

Verified 8227695d

ggml vulkan: add hardsigmoid and hardswish operations (#15762)

relent95 committed 6 days ago

Verified 0014fb4a

CUDA: Optimize `rms_norm_f32` kernel and its fused variants, giving 1-6% perf E2E (#15715)

ORippler committed 6 days ago

Verified 661ae31c

model-conversion : fix pyright errors (#15770)

danbev committed 6 days ago

Verified 407c2378

sampling : optimize dist sampler (#15704)

ggerganov committed 6 days ago

Verified cdedb70a

llama : fix incorrect model type for Gemma 270M (#15764)

danbev committed 7 days ago

Verified 2c8dac72

model-conversion : remove hardcoded /bin/bash shebangs [no ci] (#15765)

danbev committed 7 days ago

Verified 40a751ea

CANN: Add RoPE contiguous check for 310I DUP device (#15735)

hipudding committed 7 days ago

Verified 5eae9348

ggml-cpu : optimize RVV kernels (#15720)

xctan committed 7 days ago

Verified 05c0380f

model-conversion : add missing curl script [no ci] (#15761)

danbev committed 7 days ago

Verified 8c3fdf44

CANN: Mask unsupported TRANSPOSE_1D operator (#15733)

hipudding committed 7 days ago

Verified f6da8cb8

CANN: Fix type float_t to float (#15736)

noemotiovon committed 7 days ago

Verified 8a2234ea

fix: resolve unsigned int initialization warning for n_dims/size in gguf.cpp (#15754)

skrandy committed 7 days ago

Verified 3de00820

chore: Update `.clang-format` to use `BinPackArguments=true` (#15744)

ORippler committed 7 days ago

Verified 69db8a52

llama: -fa 1/0/-1 aliases for -fa on/off/auto (#15746)

JohannesGaessler committed 7 days ago

Verified c466abe1

vulkan: fix shaders gen when no integer dot is available (#15740)

0cc4m committed 7 days ago

Verified 0a2a3841

CANN: Resolve soft_max precision issue (#15730)

hipudding committed 8 days ago

Verified 9961d244

vulkan: Fix macro parameter order for f32 matmul shaders (#15716)

jeffbolznv committed 8 days ago

Verified 25f1045f

opencl: add attn sinks support for FA kernels (#15706)

rmatif committed 8 days ago

Verified 97669e40

CANN: Support eager execution mode under ACL graph compilation (#15712)

noemotiovon committed 8 days ago

Verified 2f853687

CANN: Support ext_factor in rope (#15710)

hipudding committed 8 days ago

Verified ef2af57d

ggml-backend: raise GGML_MAX_SPLIT_INPUTS (#15722)

JohannesGaessler committed 8 days ago

Verified 5d804a49

vulkan: use memory budget extension to read memory usage (#15545)

giladgd committed 8 days ago

Verified d4d8dbe3

vulkan: add missing clamps in new mul_mat_id paths (#15702)

jeffbolznv committed 8 days ago

Verified 35a42eda

Newer Older