ResearchBenchmarksGoal-conditioned Reinforcement Learning on JaxNavFollow69.2IQM (Normalized Win Rate)CPPO62.33664.11865.967.682May 13, 2026Evaluation ResultsMethodMethodLinksIQM (Normalized Win Rate)CPPONumber of tasks=1 task...Number of tasks=1 task, Action space=continuous, Agent type=multi-agent2026.0569.2ICSACNumber of tasks=1 task...Number of tasks=1 task, Action space=continuous, Agent type=multi-agent2026.0562.6