deep learning framework / tensor library
Burn is a Rust-native deep learning framework and tensor library that unifies training and inference with dynamic graphs, autodiff, and JIT kernel fusion across CPUs, GPUs (CUDA/ROCm/Metal), and WebGPU/WASM.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Choose when you want a Python-first, C++-backed DL framework with a different parallelism strategy.
OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.
Choose when you need a C++-first ML library with optimized kernels and no Python dependency.
A C++ standalone library for machine learning
Choose when you are in the JVM ecosystem and need engine-agnostic DL abstractions.
An Engine-Agnostic Deep Learning Framework in Java
Choose when you are building on JAX and want a small, functional NN library instead of a full framework.
JAX-based neural network library
Choose when you work in Nim and need a tensor library with autograd and GPU/OpenCL backends.
A fast, ergonomic and portable tensor library in Nim with a deep learning focus for CPU, GPU and embedded devices via OpenMP, Cuda and OpenCL backends
Choose when you want the dominant Python framework with a huge ecosystem and dynamic graphs.
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Choose when you need a JS/TS library with WebGL acceleration for browser-based ML.
A WebGL accelerated JavaScript library for training and deploying ML models.
Choose when you need a C++/Python DL framework with integrated training and deployment.
MegEngine 是一个快速、可拓展、易于使用且支持自动求导的深度学习框架
Choose when you work in Julia and want a flexible DL framework integrated with the language.
Relax! Flux is the ML library that doesn't make you tensor
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Choose when you want a minimal, shape-checked Rust DL framework, but expect stale maintenance.
Deep learning in Rust, with shape checked tensors and neural networks
Choose when you want a GPU library for sparse deep learning on large recommendations, despite it being archived.
Deep Scalable Sparse Tensor Network Engine (DSSTNE) is an Amazon developed library for building Deep Learning (DL) machine learning (ML) models
These projects were analysed and named burn among their alternatives. The relationship is not symmetric — how burn rates them is a separate judgement, made when burn is analysed in its own right.
calls burn “choose if you want a modern Rust deep learning framework with emphasis on flexibility and portability.”
Tensors and Dynamic neural networks in Python with strong GPU acceleration
calls burn “Choose for Rust projects that need a modern deep learning framework with eager execution and portability.”
Deep Learning for humans
calls burn “when you prefer a Rust DL framework with high portability and performance.”
A WebGL accelerated JavaScript library for training and deploying ML models.
calls burn “choose when you want a pure Rust deep learning framework with high portability”
OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.
calls burn “When you need a pure Rust deep-learning framework with portable backends and no Python runtime.”
CoreNet: A library for training deep neural networks