Skip to content
Better HN
Prism: When an LLM predicts the next token, which training does it relying on? | Better HN