chore(docs): update readme (#5080)
* WIP: Update readme * Improvements * Update README.md Co-authored-by: Guillaume Lagrange <lagrange.guillaume.1@gmail.com> * Update README.md Co-authored-by: Guillaume Lagrange <lagrange.guillaume.1@gmail.com> * Update README.md Co-authored-by: Guillaume Lagrange <lagrange.guillaume.1@gmail.com> * Fix backend comment * Preserves intuitive ergonomics --------- Co-authored-by: Guillaume Lagrange <lagrange.guillaume.1@gmail.com>
This commit is contained in:
Nathaniel Simard
@@ -13,17 +13,104 @@
|
||||
|
||||
---
|
||||
|
||||
**Burn is a next generation Tensor Library and Deep Learning Framework that doesn't compromise on
|
||||
<br /> flexibility, efficiency and portability.**
|
||||
**Burn is both a tensor library and a deep learning framework, optimized for <br /> numerical
|
||||
computing, training and inference.**
|
||||
|
||||
<br/>
|
||||
</div>
|
||||
|
||||
<div align="left">
|
||||
|
||||
Burn is both a tensor library and a deep learning framework optimized for numerical computing, model
|
||||
inference and model training. Burn leverages Rust to perform optimizations normally only available
|
||||
in static-graph frameworks, offering optimal speed without impacting flexibility.
|
||||
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
|
||||
|
||||
<div align="left">
|
||||
<img align="right" src="https://raw.githubusercontent.com/tracel-ai/burn/main/assets/ember-blazingly-fast.png" height="96px"/>
|
||||
|
||||
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.
|
||||
|
||||
</div>
|
||||
|
||||
| Category | Project | Description |
|
||||
| ------------- | ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
|
||||
| Compute | [CubeCL](https://github.com/tracel-ai/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](https://github.com/tracel-ai/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 |
|
||||
| | `burn-rl` | Reinforcement learning building blocks |
|
||||
| | `burn-dataset` | Dataset loading, transforms and ready-made sources |
|
||||
| Models | [models](https://github.com/tracel-ai/models) | Curated pre-trained models and examples built with Burn |
|
||||
| Tooling | [burn-bench](https://github.com/tracel-ai/burn-bench) | Benchmark and compare backends, tracking performance over time |
|
||||
|
||||
Burn's [CubeCL](https://github.com/tracel-ai/cubecl) backends (CUDA, ROCm, Metal, Vulkan, WebGPU,
|
||||
CPU) compose with autodiff, fusion and remote-execution decorators, while external and simpler
|
||||
backends (LibTorch and pure-Rust CPU/`no_std`) compose with autodiff only. See
|
||||
[Supported Backends](#supported-backends) below for the full matrix.
|
||||
|
||||
Every project here is open-source and actively developed. Want to help build the Rust AI ecosystem?
|
||||
The [good first issues](https://github.com/tracel-ai/burn/contribute) are a great place to start,
|
||||
and the [Contributing](#contributing) guide will get you set up.
|
||||
|
||||
<details>
|
||||
<summary>
|
||||
<b>Community crates 🌱</b>
|
||||
</summary>
|
||||
<br />
|
||||
|
||||
These crates are not maintained by Tracel, but they are part of the same Rust AI story. Anything
|
||||
that helps you load data, build environments, or ship models belongs here. Built something that
|
||||
fits? Open a PR to add it!
|
||||
|
||||
| Category | Crate | Description |
|
||||
| -------------------------- | --------------------------------------------------------------- | ----------------------------------------------------------------- |
|
||||
| Data & loading | [polars](https://github.com/pola-rs/polars) | Fast DataFrames for tabular data |
|
||||
| | [arrow-rs](https://github.com/apache/arrow-rs) | Apache Arrow columnar memory format |
|
||||
| | [image](https://github.com/image-rs/image) | Image decoding, encoding and processing |
|
||||
| | [hf-hub](https://github.com/huggingface/hf-hub) | Download models and datasets from the Hugging Face Hub |
|
||||
| Tokenization & NLP | [tokenizers](https://github.com/huggingface/tokenizers) | Fast, production-ready tokenizers |
|
||||
| | [rust-bert](https://github.com/guillaume-be/rust-bert) | Ready-to-use NLP pipelines and transformer models |
|
||||
| Numerical & linear algebra | [ndarray](https://github.com/rust-ndarray/ndarray) | N-dimensional arrays |
|
||||
| | [nalgebra](https://github.com/dimforge/nalgebra) | Linear algebra |
|
||||
| Classical ML | [linfa](https://github.com/rust-ml/linfa) | Classical ML toolkit, in the spirit of scikit-learn |
|
||||
| | [smartcore](https://github.com/smartcorelib/smartcore) | Classical ML algorithms, no BLAS/LAPACK required |
|
||||
| Inference & runtimes | [candle](https://github.com/huggingface/candle) | Minimalist ML framework with a focus on LLM inference |
|
||||
| | [mistral.rs](https://github.com/EricLBuehler/mistral.rs) | Fast, multimodal LLM inference engine |
|
||||
| | [ort](https://github.com/pykeio/ort) | ONNX Runtime bindings for hardware-accelerated inference |
|
||||
| | [tract](https://github.com/sonos/tract) | Pure-Rust inference for ONNX and NNEF models |
|
||||
| | [wonnx](https://github.com/webonnx/wonnx) | 100% Rust, WebGPU-accelerated ONNX runtime for native and the web |
|
||||
| LLM apps & RAG | [rig](https://github.com/0xPlaygrounds/rig) | Build modular LLM applications and agents |
|
||||
| | [langchain-rust](https://github.com/Abraxas-365/langchain-rust) | LangChain-style chain orchestration |
|
||||
| Embeddings & vector search | [fastembed](https://github.com/Anush008/fastembed-rs) | Generate text embeddings and rerank locally |
|
||||
| | [qdrant](https://github.com/qdrant/qdrant) | Vector search engine, written in Rust |
|
||||
| | [lancedb](https://github.com/lancedb/lancedb) | Embedded, developer-friendly vector database |
|
||||
| Computer vision | [kornia-rs](https://github.com/kornia/kornia-rs) | Low-level 3D computer vision library |
|
||||
| Simulation & environments | [rapier](https://github.com/dimforge/rapier) | Physics engine for robotics and RL environments |
|
||||
| Visualization | [rerun](https://github.com/rerun-io/rerun) | Multimodal data and CV/robotics visualization |
|
||||
| | [plotters](https://github.com/plotters-rs/plotters) | Plotting and charting |
|
||||
|
||||
</details>
|
||||
|
||||
## Backend
|
||||
|
||||
@@ -421,29 +508,28 @@ community section!
|
||||
|
||||
<details>
|
||||
<summary>
|
||||
Why use Rust for Deep Learning? 🦀
|
||||
Why use Rust for AI? 🦀
|
||||
</summary>
|
||||
<br />
|
||||
|
||||
Deep Learning is a special form of software where you need very high level abstractions as well as
|
||||
extremely fast execution time. Rust is the perfect candidate for that use case since it provides
|
||||
zero-cost abstractions to easily create neural network modules, and fine-grained control over memory
|
||||
to optimize every detail.
|
||||
to optimize every detail. To this day, the mainstream solution has been to offer APIs in Python but
|
||||
rely on bindings to low-level languages such as C/C++. This reduces portability, increases
|
||||
complexity and creates friction between researchers and engineers. Rust's approach to abstractions
|
||||
is versatile enough to tackle this two-language dichotomy, and Cargo makes it easy to build, test
|
||||
and deploy from any environment, which is usually a pain in Python.
|
||||
|
||||
It's important that a framework be easy to use at a high level so that its users can focus on
|
||||
innovating in the AI field. However, since running models relies so heavily on computations,
|
||||
performance can't be neglected.
|
||||
Rust's AI ecosystem is young, but it is real and growing quickly. Foundational pieces are already
|
||||
here: Burn and [CubeCL](https://github.com/tracel-ai/cubecl) for training and compute,
|
||||
[candle](https://github.com/huggingface/candle) for inference, Hugging Face's `tokenizers` and
|
||||
`safetensors`, and `polars` and `ndarray` for data. Betting on Rust today means betting on a stack
|
||||
that is growing, and one where contributors still shape the direction. The pieces that don't exist
|
||||
yet are opportunities rather than dead-ends (see [Contributing](#contributing)).
|
||||
|
||||
To this day, the mainstream solution to this problem has been to offer APIs in Python, but rely on
|
||||
bindings to low-level languages such as C/C++. This reduces portability, increases complexity and
|
||||
creates frictions between researchers and engineers. We feel like Rust's approach to abstractions
|
||||
makes it versatile enough to tackle this two languages dichotomy.
|
||||
|
||||
Rust also comes with the Cargo package manager, which makes it incredibly easy to build, test, and
|
||||
deploy from any environment, which is usually a pain in Python.
|
||||
|
||||
Although Rust has the reputation of being a difficult language at first, we strongly believe it
|
||||
leads to more reliable, bug-free solutions built faster (after some practice 😅)!
|
||||
Rust is also what makes one-stack-everywhere possible: a single self-contained binary with no Python
|
||||
runtime to ship, running from servers down to `no_std` embedded targets.
|
||||
|
||||
</details>
|
||||
|
||||
@@ -507,7 +593,7 @@ any background. You can ask your questions and share what you built with the com
|
||||
|
||||
<br/>
|
||||
|
||||
**Contributing**
|
||||
### Contributing
|
||||
|
||||
Before contributing, please read the [Contributing Guidelines](./CONTRIBUTING.md) and our
|
||||
[Code of Conduct](./CODE-OF-CONDUCT.md). The [Contributor Book](https://burn.dev/contributor-book/)
|
||||
|
||||
Reference in New Issue
Block a user