ResearchBenchmarksDexterous Manipulation on ShadowHandFollow13,762Average PerformanceCPO10,155.2811,091.6412,02812,964.36Mar 2, 2026Evaluation ResultsMethodMethodLinksAverage PerformanceCPOEnvironment steps=2 ×...Environment steps=2 × 10^102026.0313,762SAPGEnvironment steps=2 ×...Environment steps=2 × 10^102026.0312,882PPOEnvironment steps=2 ×...Environment steps=2 × 10^102026.0310,661PBTEnvironment steps=2 ×...Environment steps=2 × 10^102026.0310,294