{"url":"/sota/atari-games-on-atari-2600-solaris","task":{"name":"Atari Games","url":"/task/atari-games","note":null},"dataset":{"name":"Atari 2600 Solaris","url":null},"category":"Playing Games","categories":["Playing Games"],"category_note":null,"description":"The Atari 2600 Games task (and dataset) involves training an agent to achieve high game scores.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [Playing Atari with Deep Reinforcement Learning](https://arxiv.org/pdf/1312.5602v1.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Score"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Score":"higher"}},"counts":{"rows":23,"rows_with_code":20,"rows_with_paper_page":23,"rows_dated":23,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"Agent57","metrics":{"Score":"44199.93"},"uses_additional_data":false,"paper_date":"2020-03-30","paper":"/paper/agent57-outperforming-the-atari-human","paper_url":"https://arxiv.org/abs/2003.13350v1","paper_title":"Agent57: Outperforming the Atari Human Benchmark","code":"https://github.com/michaelnny/deep_rl_zoo","n_code_links":5,"syntology":{"n_ran":0,"n_unverified":9,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"Go-Explore","metrics":{"Score":"19671"},"uses_additional_data":false,"paper_date":"2020-04-27","paper":"/paper/first-return-then-explore","paper_url":"https://arxiv.org/abs/2004.12919v6","paper_title":"First return, then explore","code":"https://github.com/uber-research/go-explore","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":3,"model":"SND-STD","metrics":{"Score":"12460"},"uses_additional_data":false,"paper_date":"2023-02-22","paper":"/paper/exploration-by-self-supervised-exploitation","paper_url":"https://arxiv.org/abs/2302.11563v4","paper_title":"Self-supervised network distillation: an effective approach to exploration in sparse reward environments","code":"https://github.com/iskandor/snd","n_code_links":2,"syntology":null},{"rank_in_archive_order":4,"model":"SND-VIC","metrics":{"Score":"11865"},"uses_additional_data":false,"paper_date":"2023-02-22","paper":"/paper/exploration-by-self-supervised-exploitation","paper_url":"https://arxiv.org/abs/2302.11563v4","paper_title":"Self-supervised network distillation: an effective approach to exploration in sparse reward environments","code":"https://github.com/iskandor/snd","n_code_links":2,"syntology":null},{"rank_in_archive_order":5,"model":"SND-V","metrics":{"Score":"11582"},"uses_additional_data":false,"paper_date":"2023-02-22","paper":"/paper/exploration-by-self-supervised-exploitation","paper_url":"https://arxiv.org/abs/2302.11563v4","paper_title":"Self-supervised network distillation: an effective approach to exploration in sparse reward environments","code":"https://github.com/iskandor/snd","n_code_links":2,"syntology":null},{"rank_in_archive_order":6,"model":"GDI-I3","metrics":{"Score":"11074"},"uses_additional_data":false,"paper_date":"2021-06-11","paper":"/paper/gdi-rethinking-what-makes-reinforcement","paper_url":"https://arxiv.org/abs/2106.06232v6","paper_title":"GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":7,"model":"GDI-I3","metrics":{"Score":"11074"},"uses_additional_data":false,"paper_date":"2022-06-07","paper":"/paper/generalized-data-distribution-iteration","paper_url":"https://arxiv.org/abs/2206.03192v4","paper_title":"Generalized Data Distribution Iteration","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":8,"model":"GDI-H3","metrics":{"Score":"9105"},"uses_additional_data":false,"paper_date":"2022-06-07","paper":"/paper/generalized-data-distribution-iteration","paper_url":"https://arxiv.org/abs/2206.03192v4","paper_title":"Generalized Data Distribution Iteration","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"CGP","metrics":{"Score":"8324"},"uses_additional_data":false,"paper_date":"2018-06-14","paper":"/paper/evolving-simple-programs-for-playing-atari","paper_url":"http://arxiv.org/abs/1806.05695v1","paper_title":"Evolving simple programs for playing Atari games","code":"https://github.com/ShuhuaGao/gpFlappyBird","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":9,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":10,"model":"IQN","metrics":{"Score":"8007"},"uses_additional_data":false,"paper_date":"2018-06-14","paper":"/paper/implicit-quantile-networks-for-distributional","paper_url":"http://arxiv.org/abs/1806.06923v1","paper_title":"Implicit Quantile Networks for Distributional Reinforcement Learning","code":"https://github.com/opendilab/DI-engine/blob/main/ding/policy/iqn.py","n_code_links":19,"syntology":{"n_ran":0,"n_unverified":2,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"QR-DQN-1","metrics":{"Score":"6740"},"uses_additional_data":false,"paper_date":"2017-10-27","paper":"/paper/distributional-reinforcement-learning-with-1","paper_url":"http://arxiv.org/abs/1710.10044v1","paper_title":"Distributional Reinforcement Learning with Quantile Regression","code":"https://github.com/DLR-RM/stable-baselines3","n_code_links":17,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":1}},{"rank_in_archive_order":12,"model":"NoisyNet-Dueling","metrics":{"Score":"6522"},"uses_additional_data":false,"paper_date":"2017-06-30","paper":"/paper/noisy-networks-for-exploration","paper_url":"https://arxiv.org/abs/1706.10295v3","paper_title":"Noisy Networks for Exploration","code":"https://github.com/opendilab/DI-engine","n_code_links":15,"syntology":{"n_ran":1,"n_unverified":2,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":13,"model":"MuZero (Res2 Adam)","metrics":{"Score":"5132.95"},"uses_additional_data":false,"paper_date":"2021-04-13","paper":"/paper/online-and-offline-reinforcement-learning-by","paper_url":"https://arxiv.org/abs/2104.06294v1","paper_title":"Online and Offline Reinforcement Learning by Planning with a Learned Model","code":"https://github.com/DHDev0/Muzero-unplugged","n_code_links":2,"syntology":null},{"rank_in_archive_order":14,"model":"Advantage Learning","metrics":{"Score":"4785.16"},"uses_additional_data":false,"paper_date":"2015-12-15","paper":"/paper/increasing-the-action-gap-new-operators-for","paper_url":"http://arxiv.org/abs/1512.04860v1","paper_title":"Increasing the Action Gap: New Operators for Reinforcement Learning","code":"https://github.com/janhuenermann/neurojs","n_code_links":2,"syntology":null},{"rank_in_archive_order":15,"model":"R2D2","metrics":{"Score":"3787.2"},"uses_additional_data":false,"paper_date":"2019-05-01","paper":"/paper/recurrent-experience-replay-in-distributed","paper_url":"https://openreview.net/forum?id=r1lyTjAqYX","paper_title":"Recurrent Experience Replay in Distributed Reinforcement Learning","code":"https://github.com/opendilab/DI-engine/blob/main/ding/policy/r2d2.py","n_code_links":3,"syntology":null},{"rank_in_archive_order":16,"model":"ASL DDQN","metrics":{"Score":"3506.8"},"uses_additional_data":false,"paper_date":"2023-05-07","paper":"/paper/train-a-real-world-local-path-planner-in-one","paper_url":"https://arxiv.org/abs/2305.04180v3","paper_title":"Train a Real-world Local Path Planner in One Hour via Partially Decoupled Reinforcement Learning and Vectorized Diversity","code":"https://github.com/xinjinghao/color","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"RND","metrics":{"Score":"3282"},"uses_additional_data":false,"paper_date":"2018-10-30","paper":"/paper/exploration-by-random-network-distillation","paper_url":"http://arxiv.org/abs/1810.12894v1","paper_title":"Exploration by Random Network Distillation","code":"https://github.com/opendilab/DI-engine/blob/main/ding/reward_model/rnd_reward_model.py","n_code_links":22,"syntology":{"n_ran":26,"n_unverified":17,"n_samples":43,"n_pointer_only_licence":15}},{"rank_in_archive_order":18,"model":"Ape-X","metrics":{"Score":"2892.9"},"uses_additional_data":false,"paper_date":"2018-03-02","paper":"/paper/distributed-prioritized-experience-replay","paper_url":"http://arxiv.org/abs/1803.00933v1","paper_title":"Distributed Prioritized Experience Replay","code":"https://github.com/ray-project/ray/tree/master/rllib","n_code_links":15,"syntology":{"n_ran":0,"n_unverified":15,"n_samples":15,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"IMPALA (deep)","metrics":{"Score":"2365.00"},"uses_additional_data":false,"paper_date":"2018-02-05","paper":"/paper/impala-scalable-distributed-deep-rl-with","paper_url":"http://arxiv.org/abs/1802.01561v3","paper_title":"IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures","code":"https://github.com/ray-project/ray/tree/master/rllib","n_code_links":24,"syntology":{"n_ran":16,"n_unverified":18,"n_samples":34,"n_pointer_only_licence":3}},{"rank_in_archive_order":20,"model":"DQNMMCe","metrics":{"Score":"2244.6"},"uses_additional_data":false,"paper_date":"2018-07-31","paper":"/paper/count-based-exploration-with-the-successor","paper_url":"https://arxiv.org/abs/1807.11622v4","paper_title":"Count-Based Exploration with the Successor Representation","code":"https://github.com/mcmachado/count_based_exploration_sr","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":2,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":21,"model":"DNA","metrics":{"Score":"2225"},"uses_additional_data":false,"paper_date":"2022-06-20","paper":"/paper/dna-proximal-policy-optimization-with-a-dual","paper_url":"https://arxiv.org/abs/2206.10027v2","paper_title":"DNA: Proximal Policy Optimization with a Dual Network Architecture","code":"https://github.com/maitchison/PPO","n_code_links":1,"syntology":null},{"rank_in_archive_order":22,"model":"DreamerV2","metrics":{"Score":"922"},"uses_additional_data":false,"paper_date":"2020-10-05","paper":"/paper/mastering-atari-with-discrete-world-models-1","paper_url":"https://arxiv.org/abs/2010.02193v4","paper_title":"Mastering Atari with Discrete World Models","code":"https://github.com/opendilab/DI-engine","n_code_links":9,"syntology":{"n_ran":3,"n_unverified":12,"n_samples":15,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"MuZero","metrics":{"Score":"56.62"},"uses_additional_data":false,"paper_date":"2019-11-19","paper":"/paper/mastering-atari-go-chess-and-shogi-by","paper_url":"https://arxiv.org/abs/1911.08265v2","paper_title":"Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model","code":"https://github.com/werner-duvaud/muzero-general","n_code_links":18,"syntology":{"n_ran":43,"n_unverified":21,"n_samples":64,"n_pointer_only_licence":62}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":12,"rows_with_any_sample_ran":7,"distinct_papers_with_graph_line":12,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":93,"n_unverified":109,"n_samples":202,"n_pointer_only_licence":88,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":93,"n_unverified":109,"n_samples":202,"n_pointer_only_licence":88,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}