ResearchBenchmarksMahjong on Mahjong 100 held-out trajectories (test)Mahjong on Mahjong 100 held-out trajectories (test)Follow77.2D_policyPPO25.92839.23952.5565.861Jun 20, 2025Evaluation resultsMethodMethodLinksD_policyPPODemonstrator=Bot CDemonstrator=Bot C2025.0677.2PPODemonstrator=Bot BDemonstrator=Bot B2025.0669.1PPODemonstrator=Bot ADemonstrator=Bot A2025.0667.8MPPODemonstrator=Bot BDemonstrator=Bot B2025.0631.8MPPODemonstrator=Bot ADemonstrator=Bot A2025.0629.7MPPODemonstrator=Bot CDemonstrator=Bot C2025.0627.9