ResearchBenchmarksReinforcement Learning on Ant CrossMazeFollow1,010.04RewardDTSemNets188.6584401.9017615.145828.3883May 18, 2026Evaluation ResultsMethodMethodLinksRewardDTSemNets2026.051,010.04VIPER (PPO)2026.05996.54PPO2026.05657.03π-PRLPolicy Type=DiscretizedPolicy Type=Discretized2026.05506.32π-PRLPolicy Type=Continuous...Policy Type=Continuous relaxed2026.05363.44π-PRLPolicy Type=Fine-tunedPolicy Type=Fine-tuned2026.05223.58DiPRL2026.05220.25