A Rust-based deep learning framework and tensor library optimized for flexibility, efficiency, and cross-platform portability.
Burn is a comprehensive deep learning framework and tensor library built in Rust, designed to bridge the gap between high-level model development and low-level performance optimization. It enables seamless training and inference across diverse hardware, from embedded devices to large GPU clusters, without sacrificing flexibility.
Deep learning practitioners and engineers who need a single framework for both research and production deployment, particularly those working in Rust or requiring cross-platform compatibility from embedded systems to web browsers.
Developers choose Burn for its unique combination of Rust's performance and safety with high-level abstractions, eliminating the traditional split between research and deployment environments. Its swappable backend system with decorators like Autodiff and Fusion provides broad hardware support and automatic optimizations without code changes.
Burn is a next generation tensor library and Deep Learning Framework that doesn't compromise on flexibility, efficiency and portability.
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Supports multiple GPU backends (CUDA, ROCm, Metal, Vulkan, WebGPU) and CPU backends, enabling deployment on diverse hardware from embedded devices to large clusters without code changes.
Features like Autodiff for automatic differentiation, Fusion for kernel fusion, and Remote for distributed computations enhance performance and functionality transparently, as shown in the README examples.
Includes a terminal UI based on Ratatui for real-time monitoring of metrics, visualization, and safe interruption handling during training, improving workflow ergonomics.
Runs in WebAssembly for browser inference and supports no_std for embedded systems via the NdArray backend, ensuring seamless portability from training to deployment.
The ONNX import crate only supports a limited set of operators, restricting the types of models that can be easily ported from frameworks like TensorFlow or PyTorch.
Requires proficiency in Rust, which may deter teams accustomed to Python-based deep learning frameworks, despite Rust's benefits for performance and safety.
The project is in active development with acknowledged breaking changes, making it less stable for production use without frequent updates and potential code adjustments.
burn is an open-source alternative to the following products:
TensorFlow is an open-source machine learning framework developed by Google for building and deploying ML models across various platforms.
PyTorch is an open-source machine learning framework that provides tensor computation with strong GPU acceleration and deep neural networks built on a tape-based autograd system.