ResearchBenchmarksOffline Reinforcement Learning on HalfCheetah Medium-Replay Delayed Reward D4RLFollow42.4ScoreISCT6.41615.75825.134.442Feb 12, 2026Evaluation ResultsMethodMethodLinksScoreISCTReward Type=DelayedReward Type=Delayed2026.0242.4DTReward Type=DelayedReward Type=Delayed2026.0233QDTReward Type=DelayedReward Type=Delayed2026.0232.8CQLReward Type=DelayedReward Type=Delayed2026.027.8