Alessandro Abate

Stabilizing Policy Gradients for Sample-Efficient Reinforcement Learning in LLM Reasoning

Subversion Strategy Eval: Can LLMs statelessly strategize to subvert control protocols?

Sound Probabilistic Safety Bounds for Large Language Models

Certified Neural Approximations of Nonlinear Dynamics

Games for AI Control: Models of Safety Evaluations of AI Deployment Protocols

DeepLTL: Learning to Efficiently Satisfy Complex LTL Specifications for Multi-Task RL

Temporal-Difference Variational Continual Learning

Temporal-Difference Variational Continual Learning

The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret

The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret

Q-based Variational Inverse Reinforcement Learning

Temporal-Difference Variational Continual Learning

Networked Communication for Decentralised Agents in Mean-Field Games