This repository contains the code and journey of learning and applying the Burn deep learning framework in Rust. It's a live document of building understanding from the ground up, culminating in practical examples like MNIST digit recognition.
To clearly demonstrate and document the process of building neural networks with Burn, covering:
✓ Tensors & Automatic Differentiation
✓ Modular Network Design
✓ Data Loading & Batching Pipelines
✓ Training Loops with the Burn Learner
✓ Saving, Loading, and Inferring with Models
- Showcasing Burn: A powerful, modern deep learning toolkit for Rust.
- Rust for Speed & Safety: Leveraging Rust's strengths in the AI/ML domain.
- Practical Learning Resource: Provides working examples and a relatable learning path for newcomers.
Follow along, contribute, or use the examples to kickstart your own AI projects in Rust!
If you are curious, or want to try something please contribute to this project. You can contribute with:
- pull requests for improvement
- issue for unclear part or improvement ideas
- requests for missing area to explore.
Working together is the best aproach to learn and grow
To run the examples (once you have Rust and Cargo installed):
- Clone the repository:
git clone git@github.com:spaghetty/rust_ai_ml_adventure.git cd rust_ai_ml_adventure - To run the training for an example (e.g., from the MNIST project):
cargo run --example <name_of_the_example> ##(eg. base_tensor)
- To run inference with a trained model (specific command might vary per example):
cargo test cargo test --test test_tensor ##(eg. single file)
- Post 1: - Intro & Tensors
- Post 2: - Autodiff
- ... and more to come!