Stuart Russell

Emergence of Superposition: Unveiling the Training Dynamics of Chain of Continuous Thought

Statistical Guarantees for Offline Domain Randomization

GSM-Agent: Understanding Agentic Reasoning Using Controllable Environments

How Do Transformers Perform Two-Hop Reasoning in Context?

Monitoring Latent World States in Language Models with Propositional Probes

Reasoning by Superposition: A Theoretical Perspective on Chain of Continuous Thought

Diffusion On Syntax Trees For Program Synthesis

Extractive Structures Learned in Pretraining Enable Generalization on Finetuned Facts

Robust and Diverse Multi-Agent Learning via Rational Policy Gradient

BAMDP Shaping: a Unified Framework for Intrinsic Motivation and Reward Shaping

Generalization or Hallucination? Understanding Out-of-Context Reasoning in Transformers

RL, but don't do anything I wouldn't do

AssistanceZero: Scalably Solving Assistance Games

Avoiding Catastrophe in Online Learning by Asking for Help

Avoiding Catastrophe in Online Learning by Asking for Help

Observation Interference in Partially Observable Assistance Games

ALMANACS: A Simulatability Benchmark for Language Model Explainability