Recently, I have read the book Build a large Language Model From Scratch from book (This Book is really awesome! Strong recommend!). The book tells me how to build a simple GPT by PyTorch. It shows what is Transformer and how it works. It showed me how to prepare dataset, train a model, fine tune a model from a base model which has a huge parameters.
I also like Rust language very much. And I already starred a repo Burn that is the deep learning framework wrote by Rust. I can tell you it's PyTorch of Rust. Burn truly binds libtorch API. So it's naturally, I want to build a large Language Model From Scratch in Rust By Burn.