Gauthier Gidel

Associate Professor@Université de Montréal·加拿大·OpenReview

研究方向

Optimization · non-convex · generalization · implicit regularization · saddle point · generative modeling · Game theory · Adversarial examples · Generative Models · Generative Flow Networks · Robustness of LLM safety alignment · safety · soft reinforcement learning · multi agent RL · cooperative AI

Gauthier Gidel

Discrete Compositional Generation via General Soft Operators and Robust Reinforcement Learning

A Generative Approach to LLM Harmfulness Mitigation with Red Flag Tokens

Dimension-adapted Momentum Outscales SGD

Advantage Alignment Algorithms

Tight Lower Bounds and Improved Convergence in Performative Prediction

Solving hidden monotone variational inequalities with surrogate losses

Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety Tuning

Self-Play $Q$-Learners Can Provably Collude in the Iterated Prisoner's Dilemma