ResearchBenchmarksContinual Reinforcement Learning on CoinRun Two-cycle (train)Continual Reinforcement Learning on CoinRun Two-cycle (train)Follow0.521C1 Final ScoreDV3-0.136280.034360.2050.37564Mar 12, 2026Evaluation resultsMethodMethodLinksC1 Final ScoreC2 Final ScoreMax Final ScoreC1 Forward TransferC2 Forward TransferRecovery ScoreAccuracy ScoreMin Accuracy ScoreWC Accuracy ScoreDV3Task Order=Two-cycleTask Order=Two-cycle2026.030.5210.5280.2330.4010.6381.1590.633-0.1170.071TES-SACTask Order=Two-cycleTask Order=Two-cycle2026.030.022-0.039-0.026-0.0270.151.090.8830.4820.6ARROWTask Order=Two-cycleTask Order=Two-cycle2026.03-0.1110.099-0.0890.4010.7171.1841.3310.9330.912