影响力指数

18.23/100

前 45.8%

全站排名 #29,513

发表论文4 篇

平均评分4.6

年均产出2.0 篇/年

Rohan Subramani

PhD student@University of Toronto·加拿大·OpenReview

How does information access affect LLM monitors' ability to detect sabotage?

ICLR 2026Rejected

Password-Activated Shutdown Protocols for Misaligned Frontier Agents

ICLR 2026Rejected

On the Expressivity of Objective-Specification Formalisms in Reinforcement Learning

ICLR 2024Poster

A Causal Model of Theory-of-Mind in AI Agents

NeurIPS 2024Rejected

合作者 (12)

Francis Rhys Ward

Raja Mehta Moreno

Shubhorup Biswas

Charlie Griffin