Offline In-Context Reinforcement Learning on K2D13-500-1 (complete)