Alexander Wettig

Language models aim to improve science learning by providing personalized assistance.

SWE-agent improves LM agents' performance in software engineering tasks with a specialized interface.

Learn how task scaling laws and model ladders improve AI predictions.