I built a pure WGSL LLM engine to run Llama on my Snapdragon laptop GPU

by Beledarian | View on Hacker News