Offline Reinforcement Learning on D4RL AntMaze v0 (Aggregate of All Splits) benchmark leaderboard · SOTA2 Research

Offline Reinforcement Learning on D4RL AntMaze v0 (Aggregate of All Splits)

96.8umaze ReturnCPED

Evaluation Results

Method	Links
CPED2023.01		96.8	55.6	85.1	72.1	34.9	32.3	376.8
SPOT2023.01		93.5	40.7	74.7	79.1	35.3	36.3	359.6
IQL2023.01		89.6	65.6	76.4	72.8	42	46	392.4
CQL2023.01		82.6	10.2	59	46.6	16.4	3.2	218
BCQ2023.01		78.9	55	0	0	6.7	2.2	142.8
BEAR2023.01		73	61	0	8	0	0	142
TD3+BC2023.01		73	47	0	0.2	0	0	120.2
PLAS2023.01		62	45.4	31.4	20.6	2.2	3	164.6
DT2023.01		54.2	41.2	0	0	0	0	95.4
BC2023.01		49.2	41.8	0.4	0.2	0	0	91.6