影响力指数

39.63/100

前 14.7%

全站排名 #9,449

发表论文8 篇

平均评分5.6

年均产出2.7 篇/年

Rohin Shah

Researcher@Google DeepMind·英国·OpenReview

研究方向

ai safety · preference learning · ai alignment · interpretability

Consistency Training Helps Stop Sycophancy and Jailbreaks

ICLR 2026Rejected

Evaluating LLM Goal-Directedness

ICLR 2026Rejected

MONA: Myopic Optimization with Non-myopic Approval Can Mitigate Multi-step Reward Hacking

ICML 2025Poster

Evaluating Frontier Models for Stealth and Situational Awareness

NeurIPS 2025Rejected

合作者 (20)

Sebastian Farquhar