Pull Requests ggml-org/llama.cpp

HIP: use v_dot2_f32_f16 instruction for FA Nvidia GPU ggml

#15884 opened 2025-09-08 23:07 by JohannesGaessler

CUDA: fix GET_ROWS for large tensors Nvidia GPU ggml

#15882 opened 2025-09-08 20:58 by JohannesGaessler

contrib : add notes about merging PRs

#15881 opened 2025-09-08 18:37 by ggerganov

feat: Extra debugging support for model conversion examples python

#15877 opened 2025-09-08 11:22 by pwilkin

CUDA: Add `fastdiv` and `fastmodulo` to `k_bin_bcast*`, giving 1-3% E2E performance Nvidia GPU ggml

#15872 opened 2025-09-08 08:43 by ORippler

model-conversion : add embedding prompt file support examples python

#15871 opened 2025-09-08 08:14 by danbev

opencl: support ne3 in `get_rows` ggml OpenCL

#15866 opened 2025-09-08 06:02 by lhez

Fix loongarch lsx compilation error ggml

#15864 opened 2025-09-08 03:01 by junchao-loongson

CANN: Format ggml-cann src code using clang-format ggml Ascend NPU

#15863 opened 2025-09-08 02:50 by noemotiovon

vulkan: fix failing dequant shaders Vulkan ggml

#15862 opened 2025-09-08 02:36 by jeffbolznv

vulkan: Fix OOB accesses in soft_max_back testing Vulkan ggml

#15861 opened 2025-09-07 23:32 by jeffbolznv

llama: print memory breakdown on exit

#15860 opened 2025-09-07 21:35 by JohannesGaessler

CUDA: print CUDART_VERSION on init Nvidia GPU ggml

#15853 opened 2025-09-07 08:29 by JohannesGaessler

Feat: Apertus model implementation python ggml

#15852 opened 2025-09-07 08:14 by pwilkin

Modernize use nullptr android examples

#15851 opened 2025-09-07 08:01 by distlibs

ggml-zdnn: fix #15414, activate FP16 and BF16 acceleration and incorrect zTensor free documentation ggml IBM zDNN

#15839 opened 2025-09-06 14:15 by taronaeo

llguidance : use attrs to determine special tokens

#15837 opened 2025-09-06 13:34 by dstoc

metal : make the backend async ggml Apple Metal

#15832 opened 2025-09-06 09:28 by ggerganov

requirements : update transformers/torch for Embedding Gemma testing examples python server

#15828 opened 2025-09-06 05:04 by danbev

webgpu : fix build on emscripten build script testing ggml

#15826 opened 2025-09-05 23:22 by ngxson

Add support for Qwen3-Reranker examples python server

#15824 opened 2025-09-05 21:21 by iamlemec

Improve user-facing warnings for chat template and context length examples

#15822 opened 2025-09-05 19:37 by vinkal-chudgar

webui: fix Seed-OSS thinking block examples server

#15820 opened 2025-09-05 16:51 by ServeurpersoCom

Rewrite llama-run to use llama-server examples

#15818 opened 2025-09-05 14:06 by ericcurtin

ggml : split graph allocations according to backend max buffer size testing Vulkan ggml

#15815 opened 2025-09-05 09:42 by Acly

CANN: implement LRU cache for ACL graphs in CANN backend documentation ggml Ascend NPU

#15814 opened 2025-09-05 09:19 by noemotiovon

CUDA: Conv2d Tensor Core Nvidia GPU ggml

#15813 opened 2025-09-05 06:36 by mnehete32

Add conv2d Implicit GEMM testing Nvidia GPU ggml

#15805 opened 2025-09-04 20:16 by bssrdf

vulkan: add mul_mat variant for embedded gpus Vulkan ggml

#15800 opened 2025-09-04 16:25 by rmatif

ggml-backend : add GGML_BACKEND_DEVICE_TYPE_IGPU device type Nvidia GPU Vulkan examples ggml

#15797 opened 2025-09-04 15:14 by slaren