Tensor library for machine learning
★ 16,035⑂ 2,064👁 150⑃ 952● 587▤ 45
+255
日增长
+455
周增长
+548
月增长
16,035
stars
2,064
forks
150
watchers
952
prs
587
issues
45
releases
README
ggml Tensor library for machine learning Quick start Build from source: bash git clone https://github.com/ggml-org/ggml cd ggml mkdir build && cd build cmake .. cmake --build . --config Release -j 8 For a minimal, fully commented example (matrix multiplication), see examples/simple. Description The main goal of ggml is to be a simple, portable, and efficient tensor library for machine learning with minimal setup. - Plain C/C++ implementation without any dependencies - Cross-platform - x86, ARM, RISC-V, LoongArch, PowerPC, s390x, and WebAssembly - SIMD-optimized kernels for x86, ARM, and RISC-V - Broad backend support - CPU, GPU, NPU, and browser - 2- to 8-bit integer quantization, plus MXFP4 and NVFP4 microscaling formats - Zero memory allocations during runtime Documentation - The GGUF file format - Introduction to ggml - GGML tips & tricks Contributing - For changes to the core ggml library (including to the CMake build system), please open a PR in llama.cpp - doing so will make your PR more visible, better tested, and more likely to be reviewed
Top issues
#59Text to speech models in GGML?93 comments · 2023-04-01#8How to fine tune it?43 comments · 2022-12-09#293ggml : get rid of BLAS and all it's variants39 comments · 2023-06-25#21Suggestion: RWKV Language Model20 comments · 2023-02-06#542[Question] What is the status of Vulkan backend?17 comments · 2023-09-27#436Cannot load teknium/Replit-v2-CodeInstruct-3B14 comments · 2023-08-06#6[Feature request] Implement GPT-JT14 comments · 2022-12-01#233ggml : implement a spellcheck model (xfspell, t5-spellchecker, etc)12 comments · 2023-06-06
Discussions
#1693Transformation from Nvidia to ggml1 comments · 2026-11-14#1686clap.cpp for audio embedding0 comments · 2026-11-09#1674Optimized kernels in PTX0 comments · 2026-10-27#1649old NVIDIA GPU arch run(Fermi,Kepler) support, and LLVM-IR NVPTX(Clang) compiler support1 comments · 2026-10-04#1605Help0 comments · 2026-08-24#1579dist : releases and versioning of `ggml-org` projects17 comments · 2026-08-07#1543Verifying Q4_K scale decode: uint32 bit-op approach vs get_scale_min_k4 equivalence1 comments · 2026-06-15#1524use ggml in android NDK1 comments · 2026-05-30
Latest release
v1.0.0 v1.0.0 2026-11-13Overview ggml 1.0.0 prunes the deprecated public API and the legacy DIAG_MASK ops, adds configurable YaRN truncation and FP8 support, and brings broad performance work across backends: few-row quantized matmuls and MoE fusion on CUDA/Metal, a Vulkan prefill FlashAttention path, a device-side MoE scheduler on SYCL, decomposed Adreno FlashAttention in OpenCL, new WebGPU kernels, expanded Hexagon op coverage, and scheduler improvements that avoid uploading unused inputs. API changes - Pruned deprecated API and removed the legacy DIAG_MASK_INF/DIAG_MASK_ZERO ops, bumping the RPC protocol