Yuanfu Wang

value function (2) trajectory modeling (1) instruction following (1) reinforcement learning from human feedback (1) adversarial attack (1) reward hacking (1) decision transformer (1) inference-time alignment (1) stochastic environment (1) llm alignment (1) adversarial preference learning (1) large language model (1) return-conditioned supervised learning (1) offline reinforcement learning (1) token-level guidance (1)