Offline Reinforcement Learning on D4RL Mujoco v0 (various) benchmark leaderboard · SOTA2 Research

Offline Reinforcement Learning on D4RL Mujoco v0 (various)

35.4HalfCheetah Return (Random)MOPO

Evaluation Results

Method	Links
MOPO2021.07		35.4	11.7	13.6	42.3	17.8	28	53.1	67.5	39	63.3	44.6	23.7
CQL2021.07		35.4	10.8	7	44.4	79.2	58	46.2	48.6	26.7	62.4	98.7	111
CODAC2021.07		34.6	11	18.7	46.3	82	70.8	44.1	100.2	33.2	70.4	106	112
ORAAC2021.07		13.5	9.8	3.2	41	27.3	1.48	30	16.3	28	24	28.2	18.2
BCQ2021.07		2.2	10.6	4.9	40.7	53.1	54.5	38.2	33.1	15	64.7	57.5	110.9