game-solvers

CAST is a method that uses game solver state values to provide turn-level credit assignment signals for training LLM agents via reinforcement learning, outperforming baselines across multiple games in both in-domain and unseen-difficulty evaluations.