Yarin Gal

BiasBusters: Uncovering and Mitigating Tool Selection Bias in Large Language Models

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

FindingDory: A Benchmark to Evaluate Memory in Embodied Agents

Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs

Existing Adversarial LLM Unlearning Evaluations Are Inconclusive

RedCodeAgent: Automatic Red-teaming Agent against Diverse Code Agents

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples

OMNI-LEAK: Orchestrator Multi-Agent Network Induced Data Leakage

Deception in Dialogue: Evaluating and Mitigating Deceptive Behavior in Large Language Models

Predicting Weak-to-Strong Generalization from Latent Representations

Building Reliable Long-Form Generation via Step-Wise Hallucination Rejection Sampling

ToolTweak: An Attack on Tool Selection in LLM-based Agents

Weighted Conditional Flow Matching

Language Models Change Facts Based on the Way You Talk

Temporal-Difference Variational Continual Learning

Protriever: End-to-End Differentiable Protein Homology Search for Fitness Prediction

Memo: Training Memory-Efficient Embodied Agents with Reinforcement Learning

Fundamental Limitations in Pointwise Defences of LLM Finetuning APIs

AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents

MIP against Agent: Malicious Image Patches Hijacking Multimodal OS Agents

Temporal-Difference Variational Continual Learning

Scaling Up Active Testing to Large Language Models

Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs

Temporal-Difference Variational Continual Learning