Hi there 👋

Notes on training LLMs using data parallelism strategy

Notes on Ultra-scale Playbook - training LLM on a single GPU

Interesting things I learnt this week

Introduction and motivation