Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

DeepMind Control Suite

Benchmarks

Task NameDataset NameSOTA ResultTrend
LocomotionDeepMind Control suite Dog-Trot
Final Return938
17
Continuous ControlDeepMind Control Suite visual observations
Acrobot Swingup Score24,829
16
Continuous ControlDeepMind Control Suite (DMC)
Cheetah Run873
15
Continuous ControlDeepMind Control (DMC) Suite (1M steps)
IQM87.1
14
Walker-runDeepMind Control Suite (DMC)
Average Return818
13
Walker-walkDeepMind Control Suite (DMC)
Total Average Return988
13
Walker-standDeepMind Control Suite (DMC)
Total Return997
13
Dog-walkDeepMind Control Suite (DMC)
Total Return976
13
Dog-runDeepMind Control Suite (DMC)
Total Average Return628
13
Continuous ControlDeepMind Control Suite Cheetah Run
Reward836
13
Continuous ControlDeepMind Control Suite Reacher Hard (test)
Reward975.35
12
Continuous ControlDeepMind Control Suite Point Mass - Easy
Reward912.65
12
Hopper HopDeepMind Control Suite (DMC)
Steps Required (k)153.5
12
World model image predictionDeepMind Control Suite Humanoid
MSE4.2077
12
Physical state predictionDeepMind Control Suite Humanoid Easy tasks (random policy)
MSE0.6535
12
Physical state predictionDeepMind Control Suite Reacher Easy tasks (random policy)
MSE0.0005
12
Physical state predictionDeepMind Control Suite Cheetah Easy tasks (random policy)
MSE0.1206
12
Walker RunDeepMind control suite
Average Return856
11
Hopper HopDeepMind control suite
Average Return511
11
Walker RunDeepMind Control Suite (DMC)
Steps (k)512.6
10
Cup CatchDeepMind Control Suite (DMC)
Sample Efficiency (Steps)104,700
10
World model image predictionDeepMind Control Suite Acrobot
MSE0.1806
10
World model image predictionDeepMind Control Suite Cheetah
MSE0.1565
10
Physical state predictionDeepMind Control Suite Acrobot Easy tasks (random policy)
MSE0.0001
10
Walker WalkDeepMind Control Suite (in-distribution)
Average Return996.502
10
Showing 25 of 155 rows