tracel-ai /burn
Burn is a next generation tensor library and Deep Learning Framework that doesn't compromise on flexibility, efficiency and portability.
★ 16,280⑂ 1,128👁 92⑃ 4,624● 1,646▤ 46
+6
Daily
16,280
stars
1,128
forks
92
watchers
4,624
prs
1,646
issues
46
releases
README
A tensor library and deep learning framework for numerical computing, training and inference. --- Training and inference usually live in separate worlds. Models are typically trained in Python then exported to an open format like ONNX or optimized for production engines like vLLM, ONNX Runtime, or TensorRT. This export step is often brittle and lossy, ruling out complex architectures and advanced deployment use cases. Burn unifies the two. By executing multi-platform tensor operations via a single, unified API, the exact code used for training is the exact code that runs in production. This makes workloads like on-device personalization and federated learning straightforward, while enabling teams to go from prototype to deployment in a single codebase. Burn preserves the intuitive ergonomics of PyTorch, with dynamic shapes and graphs, but JIT-compiles streams of tensor operations, performing automatic kernel fusion. You get the flexibility of dynamic graphs without the performance drop. Rust for Research? Rust used to be a tough sell for research: long compilation times disrupted the fast edit-compile-run loop that draws researchers to Python. Burn changes this paradigm. Designed around incremental compilation, modifying model code recompiles in under 5 seconds, even in release mode. This delivers a Python-like feedback loop with the speed and safety of Rust. Ecosystem Burn is the core of a growing, fully open-source Rust AI ecosystem. You are not adopting a single library, you are joining a stack that spans GPU compute, model interop and domain toolkits, with plenty of room to help shape what comes next. | Category | Project | Description | | ------------- | ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Compute | CubeCL | GPU compute language and compiler behind Burn's accelerated backends. Write kernels once in Rust, run on CUDA, ROCm, Metal, Vulkan and WebGPU. Usable standalone. | | Model interop | burn-onnx | Import ONNX models into Burn as native Rust code | | | burn-store | Save, load and import model weights, including PyTorch and Safetensors | | Domains | burn-vision | Computer vision operators and building blocks | |
Top issues
#1187store: import weights from GGUF files24 comments · 2024-01-29#1942autodiff: support higher-order derivatives23 comments · 2024-06-29#4512metal: performance comparison with MLX on Qwen3-0.6B20 comments · 2026-02-12#544train: add GPU utilization and memory metrics for non-NVIDIA GPUs20 comments · 2023-07-25#5891wgpu: system memory grows over multi-day training runs19 comments · 2026-09-28#3624tensor: design dynamic rank tensors12 comments · 2025-08-27#851examples: replace the web example shell scripts with a cross-platform build tool12 comments · 2023-10-06#846tensor: add sparse tensor support12 comments · 2023-10-04
Discussions
#6489Does burn-onnx support npu for backend?0 comments · 2026-11-11#6436Accelerate.framework (OpenBLAS) backend?0 comments · 2026-11-07#6422What is the role of CubeK?1 comments · 2026-11-06#6264Train/Test/Validate Split in TUI?1 comments · 2026-10-28#6185Plans for the new Burn AI engine (after burn-lm)? Looking to contribute: serving, training, systems work1 comments · 2026-10-22#6030Compile-time kernel customization for transparent operation overrides1 comments · 2026-10-10#5826Q-ANT inference backend1 comments · 2026-09-24#5660Sharing a fuzzer for Burn1 comments · 2026-09-13
Latest release
v0.22.1 v0.22.1 2026-10-20What's Changed - fix(fusion): keep an indexed op's output as its block's reference (5991) @nathanielsimard - fix(module): clarify derive field handling and suggest [module(skip)] (5997) @laggui - fix(derive): preserve generic submodules with primitive type arguments (6000) @laggui - feat: expose persistence feature for CubeCL backends (6012) @laggui - fix(docs): enable branches std workaround and repair optional API links (6016) @laggui - fix(dataset): enable branches std workaround for all sqlite builds (6092) @getong - fix(deps): update rustls to 0.23.46 (6070) @laggui - docs(wasm)