List your company

ResearchFormerly WizWand

DMControl

Benchmarks

TaskDatasetSOTA resultTrendResults

Continuous Control

DMControl 500k

979Spin Score

42

Continuous Control

DMControl 100k

986.38DMControl: Finger Spin Score

38

DMControl

916.9DMControl: Ball in Cup Catch Score

29

Visual Reinforcement Learning

DMControl Ball in cup, Catch

960Episode Return

16

Visual Reinforcement Learning

DMControl Finger, Spin

976Episode Return

16

Visual Reinforcement Learning

DMControl Walker Walk

802Episode Return

16

Visual Reinforcement Learning

DMControl Cheetah Run

504Episode Return

16

Visual Reinforcement Learning

DMControl Reacher Easy

969Episode Return

16

Visual Reinforcement Learning

DMControl Cartpole, Swingup

872Episode Return

16

Reinforcement Learning

DMControl

0.024Hopper/Hop Error

13

DMControl Novel view (test)

917.21Reward

12

DMControl Novel view (test)

973.6Reward

12

Offline Reinforcement Learning

DMControl walker-walk (expert)

97.97Normalized Score

12

Offline Reinforcement Learning

DMControl cheetah-run (expert)

96.28Normalized Score

12

Visual Reinforcement Learning

DMControl-GB Video-Easy

871Walker Walk Score

10

Vision-based Reinforcement Learning

DMControl 500k

983Finger Spin Score

8

Vision-based Reinforcement Learning

DMControl 100k

942Finger Spin Score

8

DMControl Hopper

0.81Average Normalized Return

8

DMControl HalfCheetah

72Average Normalized Return

8

Visual Reinforcement Learning

DMControl VDCS Markov-temporal perturbations (test)

864Cartpole Swingup Score

8

Continuous Control

DMControl Novel view

770.56Episode Reward

8

Continuous Control

DMControl

885Point Mass Easy

7

Zero-shot Continuous Control

DMControl-GB average generalization

808Cartpole Swingup Score

7

Continuous Control

DMControl GB (train)

872Cartpole Swingup Score

7

DMControl-GB color-easy (test)

346Average Episode Return

7