paper-with-me

Atari Games 벤치마크

Atari Games on Atari 2600 Freeway

118개 결과 · ⬇ CSV · JSON

Score

0 8.5 17 25.5 34 2012-07 2026-11 Best Baseline — 22.5 (2012-07-19) Best Learner — 19.1 (2012-07-19) UCT — 0.4 (2012-07-19) Best Baseline — 22.5 (2012-07-19) Best Learner — 19.1 (2012-07-19) UCT — 0.4 (2012-07-19) Nature DQN — 30.3 (2015-02-25) Nature DQN — 30.3 (2015-02-25) MP-EB — 27.0 (2015-07-03) MP-EB — 27.0 (2015-07-03) Gorila — 10.2 (2015-07-15) Gorila — 10.2 (2015-07-15) DQN noop — 30.8 (2015-09-22) DDQN (tuned) hs — 28.8 (2015-09-22) Prior+Duel hs — 28.2 (2015-09-22) DQN hs — 26.9 (2015-09-22) DQN noop — 30.8 (2015-09-22) DDQN (tuned) hs — 28.8 (2015-09-22) Prior+Duel hs — 28.2 (2015-09-22) DQN hs — 26.9 (2015-09-22) Prior noop — 33.7 (2015-11-18) Prior hs — 28.9 (2015-11-18) Prior noop — 33.7 (2015-11-18) Prior hs — 28.9 (2015-11-18) DDQN (tuned) noop — 33.3 (2015-11-20) Prior+Duel noop — 33.0 (2015-11-20) Duel hs — 0.2 (2015-11-20) Duel noop — 0.0 (2015-11-20) DDQN (tuned) noop — 33.3 (2015-11-20) Prior+Duel noop — 33.0 (2015-11-20) Duel hs — 0.2 (2015-11-20) Duel noop — 0.0 (2015-11-20) Persistent AL — 32.3 (2015-12-15) Advantage Learning — 31.72 (2015-12-15) Persistent AL — 32.3 (2015-12-15) Advantage Learning — 31.72 (2015-12-15) A3C FF (1 day) hs — 0.1 (2016-02-04) A3C FF hs — 0.1 (2016-02-04) A3C LSTM hs — 0.1 (2016-02-04) A3C FF (1 day) hs — 0.1 (2016-02-04) A3C FF hs — 0.1 (2016-02-04) A3C LSTM hs — 0.1 (2016-02-04) Bootstrapped DQN — 33.9 (2016-02-15) Bootstrapped DQN — 33.9 (2016-02-15) DDQN+Pop-Art noop — 33.4 (2016-02-24) DDQN+Pop-Art noop — 33.4 (2016-02-24) A3C-CTS — 30.48 (2016-06-06) A3C-CTS — 30.48 (2016-06-06) TRPO-hash — 34.0 (2016-11-15) TRPO-hash — 34.0 (2016-11-15) DQN-CTS — 33.0 (2017-03-03) DQN-PixelCNN — 31.7 (2017-03-03) DQN-CTS — 33.0 (2017-03-03) DQN-PixelCNN — 31.7 (2017-03-03) ES FF (1 hour) noop — 31.0 (2017-03-10) ES FF (1 hour) noop — 31.0 (2017-03-10) Sarsa-ε — 29.9 (2017-06-25) Sarsa-φ-EB — 0.0 (2017-06-25) Sarsa-ε — 29.9 (2017-06-25) Sarsa-φ-EB — 0.0 (2017-06-25) NoisyNet-Dueling — 34.0 (2017-06-30) NoisyNet-Dueling — 34.0 (2017-06-30) C51 noop — 33.9 (2017-07-21) C51 noop — 33.9 (2017-07-21) QR-DQN-1 — 34.0 (2017-10-27) QR-DQN-1 — 34.0 (2017-10-27) IMPALA (deep) — 0.0 (2018-02-05) IMPALA (deep) — 0.0 (2018-02-05) Ape-X — 33.7 (2018-03-02) Ape-X — 33.7 (2018-03-02) IQN — 34.0 (2018-06-14) A2C + SIL — 32.2 (2018-06-14) CGP — 28.2 (2018-06-14) IQN — 34.0 (2018-06-14) A2C + SIL — 32.2 (2018-06-14) CGP — 28.2 (2018-06-14) POP3D — 21.21 (2018-07-02) POP3D — 21.21 (2018-07-02) DQNMMCe — 29.5 (2018-07-31) DQNMMCe — 29.5 (2018-07-31) Intrinsic Reward Agent — 32.8 (2018-08-13) Intrinsic Reward Agent — 32.8 (2018-08-13) R2D2 — 32.5 (2019-05-01) R2D2 — 32.5 (2019-05-01) SAC — 4.4 (2019-10-16) SAC — 4.4 (2019-10-16) MuZero — 33.03 (2019-11-19) MuZero — 33.03 (2019-11-19) Agent57 — 32.59 (2020-03-30) Agent57 — 32.59 (2020-03-30) CURL — 27.9 (2020-04-08) CURL — 27.9 (2020-04-08) Go-Explore — 34.0 (2020-04-27) Go-Explore — 34.0 (2020-04-27) DreamerV2 — 33.0 (2020-10-05) DreamerV2 — 33.0 (2020-10-05) Discrete Latent Space World Model (VQ-VAE) — 29.0 (2020-10-12) Discrete Latent Space World Model (VQ-VAE) — 29.0 (2020-10-12) ENAS — 22.0 (2020-11-30) SPOS — 22.0 (2020-11-30) ENAS — 22.0 (2020-11-30) SPOS — 22.0 (2020-11-30) MuZero (Res2 Adam) — 33.87 (2021-04-13) MuZero (Res2 Adam) — 33.87 (2021-04-13) GDI-I3 — 34.0 (2021-06-11) GDI-I3 — 34.0 (2021-06-11) GDI-I3 — 34.0 (2022-06-07) GDI-H3(200M frames) — 34.0 (2022-06-07) GDI-H3 — 34.0 (2022-06-07) GDI-I3 — 34.0 (2022-06-07) GDI-H3(200M frames) — 34.0 (2022-06-07) GDI-H3 — 34.0 (2022-06-07) DNA — 33.0 (2022-06-20) DNA — 33.0 (2022-06-20) ASL DDQN — 33.9 (2023-05-07) ASL DDQN — 33.9 (2023-05-07) Best Baseline — 22.5 (2012-07-19) Nature DQN — 30.3 (2015-02-25) DQN noop — 30.8 (2015-09-22) Prior noop — 33.7 (2015-11-18) Bootstrapped DQN — 33.9 (2016-02-15) TRPO-hash — 34.0 (2016-11-15)
RankModel Score PaperCodeYear
1 TRPO-hash 34.0 #Exploration: A Study of Count-Based Exploration for Deep Reinforcement Learning uoe-agents/derl · nhynes/abc · clementbernardd/Count-Based-Exploration 2016
1 IQN 34 Implicit Quantile Networks for Distributional Reinforcement Learning opendilab/DI-engine · chainer/chainerrl · Kchu/DeepRL_CK · +16 2018
1 NoisyNet-Dueling 34 Noisy Networks for Exploration opendilab/DI-engine · Curt-Park/rainbow-is-all-you-need · chainer/chainerrl · +12 2017
1 QR-DQN-1 34 Distributional Reinforcement Learning with Quantile Regression DLR-RM/stable-baselines3 · facebookresearch/Horizon · facebookresearch/ReAgent · +14 2017
1 Go-Explore 34 First return, then explore uber-research/go-explore · qgallouedec/lge 2020
1 GDI-I3 34 Generalized Data Distribution Iteration 2022
1 GDI-I3 34 GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning 2021
1 GDI-H3(200M frames) 34 Generalized Data Distribution Iteration 2022
1 GDI-H3 34 Generalized Data Distribution Iteration 2022
10 C51 noop 33.9 A Distributional Perspective on Reinforcement Learning facebookresearch/Horizon · facebookresearch/ReAgent · opendilab/DI-engine · +19 2017
10 Bootstrapped DQN 33.9 Deep Exploration via Bootstrapped DQN tensorflow/models · tensorflow/models · NervanaSystems/coach · +3 2016
10 ASL DDQN 33.9 Train a Real-world Local Path Planner in One Hour via Partially Decoupled Reinforcement Learning and Vectorized Diversity xinjinghao/color 2023
13 MuZero (Res2 Adam) 33.87 Online and Offline Reinforcement Learning by Planning with a Learned Model DHDev0/Muzero-unplugged · enpasos/muzero 2021
14 Prior noop 33.7 Prioritized Experience Replay labmlai/annotated_deep_learning_paper_implementations · hill-a/stable-baselines · NervanaSystems/coach · +74 2015
14 Ape-X 33.7 Distributed Prioritized Experience Replay ray-project/ray · vwxyzjn/cleanrl · opendilab/DI-engine · +12 2018
16 DDQN+Pop-Art noop 33.4 Learning values across many orders of magnitude 2016
17 DDQN (tuned) noop 33.3 Dueling Network Architectures for Deep Reinforcement Learning labmlai/annotated_deep_learning_paper_implementations · tensorpack/tensorpack · facebookresearch/Horizon · +70 2015
18 MuZero 33.03 Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model werner-duvaud/muzero-general · opendilab/LightZero · koulanurag/muzero-pytorch · +15 2019
19 Prior+Duel noop 33.0 Dueling Network Architectures for Deep Reinforcement Learning labmlai/annotated_deep_learning_paper_implementations · tensorpack/tensorpack · facebookresearch/Horizon · +70 2015
19 DQN-CTS 33.0 Count-Based Exploration with Neural Density Models nolisten/erl 2017
1–20 / 118 다음 → 페이지당 10 20 50 100