ResearchBenchmarksReinforcement Learning on Atari 49 gamesFollow138.7Median Human-Normalized ScoreA2C+SIL43.12467.93792.75117.563Jun 14, 2018Evaluation ResultsMethodMethodLinksMedian Human-Normalized ScoreGames Above Human ThresholdA2C+SILtraining steps=50M, fr...training steps=50M, frames=200M2018.06138.729A2Ctraining steps=50M, fr...training steps=50M, frames=200M2018.0696.123ACPERtraining steps=50M, fr...training steps=50M, frames=200M, description=A2C with prioritized replay using ACER objective2018.0646.818