Atari Games On Atari 2600 Berzerk

评估指标

Score

评测结果

各个模型在此基准测试上的表现结果

模型名称	Score	Paper Title	Repository
FQF	12422.2	Fully Parameterized Quantile Function for Distributional Reinforcement Learning
A3C FF (1 day) hs	1433.4	Asynchronous Methods for Deep Reinforcement Learning
Ape-X	57196.7	Distributed Prioritized Experience Replay
IQN	1053	Implicit Quantile Networks for Distributional Reinforcement Learning
DDQN+Pop-Art noop	1199.6	Learning values across many orders of magnitude	-
DDQN (tuned) hs	1011.1	Deep Reinforcement Learning with Double Q-learning
A3C FF hs	817.9	Asynchronous Methods for Deep Reinforcement Learning
A3C LSTM hs	862.2	Asynchronous Methods for Deep Reinforcement Learning
ES FF (1 hour) noop	686.0	Evolution Strategies as a Scalable Alternative to Reinforcement Learning
Prior noop	1305.6	Prioritized Experience Replay
DQN noop	585.6	Deep Reinforcement Learning with Double Q-learning
GDI-I3	7607	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	-
Prior+Duel hs	2178.6	Deep Reinforcement Learning with Double Q-learning
Prior hs	865.9	Prioritized Experience Replay
IMPALA (deep)	1852.70	IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
C51 noop	1645.0	A Distributional Perspective on Reinforcement Learning
Reactor 500M	2303.1	The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning	-
Agent57	61507.83	Agent57: Outperforming the Atari Human Benchmark
MuZero	85932.60	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model
DQN hs	493.4	Deep Reinforcement Learning with Double Q-learning

0 of 39 row(s) selected.