Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
Language model tokenization at GB/s