ResearchBenchmarksCombined Generalization on MiniWorld (Case 2)Follow9.2TGT Score (Case 2)FT (CONV.)-0.3682.1164.67.084Apr 9, 2026Evaluation ResultsMethodMethodLinksTGT Score (Case 2)DSTR Score (Case 2)FT (CONV.)Algorithm=PPO, Trainin...Algorithm=PPO, Training=Fully converged2026.049.240ASPECT2026.048.84FT (LIM.)Algorithm=PPO, Trainin...Algorithm=PPO, Training=Limited (20K steps)2026.047.640SOURCEAlgorithm=PPOAlgorithm=PPO2026.043.220SF-SIMP.2026.0400SF-REC.2026.0400