Lawrence Chan

Researcher@Model Evaluation and Threat Research·美国·OpenReview

研究方向

Mechanistic interpretability · reverse engineering · interpretability · Adversarial training · language models · Inverse reinforcement learning · value learning · human robot interaction

7.8

Measuring AI Ability to Complete Long Software Tasks

NeurIPS 2025Poster

通讯

7.0

RE-Bench: Evaluating Frontier AI R&D Capabilities of Language Model Agents against Human Experts

ICML 2025Spotlight

5.5

Modular addition without black-boxes: Compressing explanations of MLPs that compute numerical integration

ICLR 2025Rejected

三作

6.3

Compact Proofs of Model Performance via Mechanistic Interpretability

NeurIPS 2024Poster

通讯

5.3

The Alignment Problem from a Deep Learning Perspective

合作者 (20)