paper-with-me

Atari Games 벤치마크

Atari Games on Atari 2600 Bowling

88개 결과 · ⬇ CSV · JSON

Score

17.6 78.23 138.9 199.5 260.1 2012-07 2026-11 Best Learner — 43.9 (2012-07-19) Best Learner — 43.9 (2012-07-19) Nature DQN — 42.4 (2015-02-25) Nature DQN — 42.4 (2015-02-25) Gorila — 54.0 (2015-07-15) Gorila — 54.0 (2015-07-15) DDQN (tuned) hs — 69.6 (2015-09-22) DQN hs — 56.5 (2015-09-22) DQN noop — 50.4 (2015-09-22) Prior+Duel hs — 50.4 (2015-09-22) DDQN (tuned) hs — 69.6 (2015-09-22) DQN hs — 56.5 (2015-09-22) DQN noop — 50.4 (2015-09-22) Prior+Duel hs — 50.4 (2015-09-22) Prior hs — 52.0 (2015-11-18) Prior noop — 47.9 (2015-11-18) Prior hs — 52.0 (2015-11-18) Prior noop — 47.9 (2015-11-18) DDQN (tuned) noop — 68.1 (2015-11-20) Duel hs — 65.7 (2015-11-20) Duel noop — 65.5 (2015-11-20) Prior+Duel noop — 46.7 (2015-11-20) DDQN (tuned) noop — 68.1 (2015-11-20) Duel hs — 65.7 (2015-11-20) Duel noop — 65.5 (2015-11-20) Prior+Duel noop — 46.7 (2015-11-20) Persistent AL — 71.59 (2015-12-15) Advantage Learning — 57.41 (2015-12-15) Persistent AL — 71.59 (2015-12-15) Advantage Learning — 57.41 (2015-12-15) A3C LSTM hs — 41.8 (2016-02-04) A3C FF (1 day) hs — 36.2 (2016-02-04) A3C FF hs — 35.1 (2016-02-04) A3C LSTM hs — 41.8 (2016-02-04) A3C FF (1 day) hs — 36.2 (2016-02-04) A3C FF hs — 35.1 (2016-02-04) Bootstrapped DQN — 60.2 (2016-02-15) Bootstrapped DQN — 60.2 (2016-02-15) DDQN+Pop-Art noop — 102.1 (2016-02-24) DDQN+Pop-Art noop — 102.1 (2016-02-24) ES FF (1 hour) noop — 30.0 (2017-03-10) ES FF (1 hour) noop — 30.0 (2017-03-10) Reactor 500M — 81.0 (2017-04-15) Reactor 500M — 81.0 (2017-04-15) C51 noop — 81.8 (2017-07-21) C51 noop — 81.8 (2017-07-21) QR-DQN-1 — 77.2 (2017-10-27) QR-DQN-1 — 77.2 (2017-10-27) IMPALA (deep) — 59.92 (2018-02-05) IMPALA (deep) — 59.92 (2018-02-05) Ape-X — 17.6 (2018-03-02) Ape-X — 17.6 (2018-03-02) IQN — 86.5 (2018-06-14) CGP — 85.8 (2018-06-14) A2C + SIL — 31.1 (2018-06-14) IQN — 86.5 (2018-06-14) CGP — 85.8 (2018-06-14) A2C + SIL — 31.1 (2018-06-14) RUDDER — 179.0 (2018-06-20) RUDDER — 179.0 (2018-06-20) POP3D — 38.99 (2018-07-02) POP3D — 38.99 (2018-07-02) R2D2 — 219.5 (2019-05-01) R2D2 — 219.5 (2019-05-01) FQF — 102.3 (2019-11-05) FQF — 102.3 (2019-11-05) MuZero — 260.13 (2019-11-19) MuZero — 260.13 (2019-11-19) Agent57 — 251.18 (2020-03-30) Agent57 — 251.18 (2020-03-30) Go-Explore — 260.0 (2020-04-27) Go-Explore — 260.0 (2020-04-27) DreamerV2 — 49.0 (2020-10-05) DreamerV2 — 49.0 (2020-10-05) MuZero (Res2 Adam) — 131.65 (2021-04-13) MuZero (Res2 Adam) — 131.65 (2021-04-13) GDI-I3 — 201.9 (2021-06-11) GDI-I3 — 201.9 (2021-06-11) GDI-H3 — 205.2 (2022-06-07) GDI-I3 — 201.9 (2022-06-07) GDI-H3 — 205.2 (2022-06-07) GDI-I3 — 201.9 (2022-06-07) DNA — 181.0 (2022-06-20) DNA — 181.0 (2022-06-20) ASL DDQN — 62.4 (2023-05-07) ASL DDQN — 62.4 (2023-05-07) Best Learner — 43.9 (2012-07-19) Gorila — 54.0 (2015-07-15) DDQN (tuned) hs — 69.6 (2015-09-22) Persistent AL — 71.59 (2015-12-15) DDQN+Pop-Art noop — 102.1 (2016-02-24) RUDDER — 179.0 (2018-06-20) R2D2 — 219.5 (2019-05-01) MuZero — 260.13 (2019-11-19)
RankModel Score PaperCodeYear
1 MuZero 260.13 Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model werner-duvaud/muzero-general · opendilab/LightZero · koulanurag/muzero-pytorch · +15 2019
2 Go-Explore 260 First return, then explore uber-research/go-explore · qgallouedec/lge 2020
3 Agent57 251.18 Agent57: Outperforming the Atari Human Benchmark michaelnny/deep_rl_zoo · pocokhc/agent57 · yuta0821/agent57_pytorch · +2 2020
4 R2D2 219.5 Recurrent Experience Replay in Distributed Reinforcement Learning opendilab/DI-engine · michaelnny/deep_rl_zoo · garymm/earl 2019
5 GDI-H3 205.2 Generalized Data Distribution Iteration 2022
6 GDI-I3 201.9 GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning 2021
6 GDI-I3 201.9 Generalized Data Distribution Iteration 2022
8 DNA 181 DNA: Proximal Policy Optimization with a Dual Network Architecture maitchison/PPO 2022
9 RUDDER 179 RUDDER: Return Decomposition for Delayed Rewards ml-jku/baselines-rudder · ml-jku/rudder 2018
10 MuZero (Res2 Adam) 131.65 Online and Offline Reinforcement Learning by Planning with a Learned Model DHDev0/Muzero-unplugged · enpasos/muzero 2021
11 FQF 102.3 Fully Parameterized Quantile Function for Distributional Reinforcement Learning opendilab/DI-engine · ku2482/fqf-iqn-qrdqn.pytorch · ku2482/rljax · +3 2019
12 DDQN+Pop-Art noop 102.1 Learning values across many orders of magnitude 2016
13 IQN 86.5 Implicit Quantile Networks for Distributional Reinforcement Learning opendilab/DI-engine · chainer/chainerrl · Kchu/DeepRL_CK · +16 2018
14 CGP 85.8 Evolving simple programs for playing Atari games ShuhuaGao/gpFlappyBird · JacobLaney/cgp-tetris 2018
15 C51 noop 81.8 A Distributional Perspective on Reinforcement Learning facebookresearch/Horizon · facebookresearch/ReAgent · opendilab/DI-engine · +19 2017
16 Reactor 500M 81.0 The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning 2017
17 QR-DQN-1 77.2 Distributional Reinforcement Learning with Quantile Regression DLR-RM/stable-baselines3 · facebookresearch/Horizon · facebookresearch/ReAgent · +14 2017
18 Persistent AL 71.59 Increasing the Action Gap: New Operators for Reinforcement Learning janhuenermann/neurojs · chainer/chainerrl 2015
19 DDQN (tuned) hs 69.6 Deep Reinforcement Learning with Double Q-learning labmlai/annotated_deep_learning_paper_implementations · tensorpack/tensorpack · hill-a/stable-baselines · +94 2015
20 DDQN (tuned) noop 68.1 Dueling Network Architectures for Deep Reinforcement Learning labmlai/annotated_deep_learning_paper_implementations · tensorpack/tensorpack · facebookresearch/Horizon · +70 2015
1–20 / 88 다음 → 페이지당 10 20 50 100