Fetching the paper…

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families · Around