Benchmarks

Task NameDataset NameSOTA ResultTrendResults

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah fore thigh

8,347Average Score

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah fore shin

4,373Average Evaluation Score

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah fore foot

6,054Average Evaluation Score

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah back thigh

7,460Avg Score

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah back shin

7,176Average Return

Multi-Agent Reinforcement Learning

Multi-Agent MuJoCo HalfCheetah back foot

5,646Average Score

Offline Multi-Agent Reinforcement Learning

Multi-agent MuJoCo Swimmer (e, m1, m2, e-m1, e-m2, m1-m2)

430.7Return

Offline Multi-Agent Reinforcement Learning

Multi-agent MuJoCo HalfCheetah expert, medium, medium-replay, medium-expert

5,413.7Return

Offline Multi-Agent Reinforcement Learning

Multi-agent MuJoCo Ant expert, medium, medium-replay, medium-expert

2,162.8Return

Robotic Control

Multi-agent MuJoCo HalfCheetah (medium-expert)

3,543.7Average Return

Robotic Control

Multi-agent MuJoCo HalfCheetah (medium-replay)

2,504.7Average Return

Robotic Control

Multi-agent MuJoCo HalfCheetah (medium)

3,608.13Average Return

Robotic Control

Multi-agent MuJoCo HalfCheetah (expert)

3,383.61Average Return

Robotic Control

Multi-agent MuJoCo Ant (medium-expert)

1,720.33Average Return

Robotic Control

Multi-agent MuJoCo Ant (medium-replay)

1,105.13Avg Return

Robotic Control

Multi-agent MuJoCo Ant (medium)

1,418.44Avg Return

Robotic Control

Multi-agent MuJoCo Ant (expert)

2,055.46Average Return

Robotic Control

Multi-agent MuJoCo Hopper (medium-expert)

709Average Return

Robotic Control

Multi-agent MuJoCo Hopper (medium-replay)

774.18Average Return

Offline Multi-Agent Reinforcement Learning

Multi-agent MuJoCo HalfCheetah k=1 (e, m1, m2, e-m1, e-m2, m1-m2)

3,760.5Return