A high-throughput and memory-efficient inference and serving engine for LLMs
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step