D4RL
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
D4RL Hopper v2 (test)
776.7Mean Episode Reward
16
D4RL Walker-medium-expert v2
113Normalized Return
16
D4RL Gym
35.6Return (Hopper, Random)
16
D4RL Antmaze
91Avg Normalized Return
15
D4RL Locomotion random
53.1Avg Normalized Return
15
D4RL MuJoCo
1,020.7Total Score
15
D4RL Medium-Expert v0
97.9Average Normalized Score
15
Stochastic-D4RL HalfCheetah (medium-expert)
5,659.4Mean Return
14
D4RL Kitchen-Mixed
75Normalized Score
14
D4RL ant medium-expert
120.3Normalized Score
14
D4RL ant-medium-replay
69.6Normalized Score
14
D4RL MuJoCo and Maze2D online fine-tuning v2 v0
98Normalized Return
14
D4RL Maze2d-umaze
148.6Normalized Return
14
D4RL MuJoCo Halfcheetah-e v2 (expert)
95.6Normalized Score
14
D4RL MuJoCo walker2d-medium-replay v2
94.4Normalized Score
13
D4RL MuJoCo halfcheetah-medium-replay v2
47.7Normalized Score
13
D4RL walker2d random
37.3Normalized Score
13
D4RL halfcheetah random
38.8Normalized Score
13
D4RL mujoco-halfcheetah (med-replay)
68.2Normalized Score
13
D4RL mujoco-halfcheetah (med)
69.5Normalized Score
13
D4RL mujoco-hopper (med)
95.4Normalized Score
13
D4RL antmaze-med (play)
40.7Normalized Score
13
D4RL antmaze-umaze (fixed)
82Normalized Score
13
D4RL maze2d-large dense
133.8Normalized Score
13
D4RL maze2d-umaze (dense)
98.3Normalized Score
13