ggerganov/llama.cpp

Pull Requests Commits

Add docker:// protocol support for llama-server model pulling

ericcurtin committed 18 hours ago

ab246cba

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

jeffbolznv committed 1 day ago

Verified 0fce7a12

vulkan : update ggml_vk_instance_validation_ext_available (#15666)

danbev committed 1 day ago

Verified 8227695d

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

relent95 committed 1 day ago

Verified 0014fb4a

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

ORippler committed 1 day ago

Verified 661ae31c

model-conversion : fix pyright errors (#15770)

danbev committed 1 day ago

Verified 407c2378

sampling : optimize dist sampler (#15704)

ggerganov committed 1 day ago

Verified cdedb70a

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

danbev committed 1 day ago

Verified 2c8dac72

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

danbev committed 1 day ago

Verified 40a751ea

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

hipudding committed 1 day ago

Verified 5eae9348

ggml-cpu : optimize RVV kernels (#15720)

xctan committed 2 days ago

Verified 05c0380f

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

danbev committed 2 days ago

Verified 8c3fdf44

CANN: Mask unsupported TRANSPOSE_1D operator (#15733)

hipudding committed 2 days ago

Verified f6da8cb8

CANN: Fix type float_t to float (#15736)

noemotiovon committed 2 days ago

Verified 8a2234ea

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

skrandy committed 2 days ago

Verified 3de00820

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

ORippler committed 2 days ago

Verified 69db8a52

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

JohannesGaessler committed 2 days ago

Verified c466abe1

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

0cc4m committed 2 days ago

Verified 0a2a3841

CANN: Resolve soft_max precision issue (#15730)

hipudding committed 2 days ago

Verified 9961d244

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

jeffbolznv committed 3 days ago

Verified 25f1045f

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

rmatif committed 3 days ago

Verified 97669e40

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

noemotiovon committed 3 days ago

Verified 2f853687

CANN: Support ext_factor in rope (#15710)

hipudding committed 3 days ago

Verified ef2af57d

ggml-backend: raise GGML_MAX_SPLIT_INPUTS (#15722)

JohannesGaessler committed 3 days ago

Verified 5d804a49

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

giladgd committed 3 days ago

Verified d4d8dbe3

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

jeffbolznv committed 3 days ago

Verified 35a42eda

vulkan: disable large mmv subgroups on older Nvidia GPUs (#15717)

0cc4m committed 3 days ago

Verified fec7911f

ggml: SVE support for exponential functions (#15145)

s-goto-11 committed 3 days ago

Verified 078ce23e

ggml: aarch64: Implement SVE F16 kernels for vector functions (#15115)

Vithulep committed 3 days ago

Verified a0c2b207

convert : remove redundant code (#15708)

DamonFool committed 3 days ago

Verified 4b20d8b7

Older