Keep-Away (2v2) on Multi-agent particle environment (MPE) (test)
52.14Mean Episode Extrinsic RewardHAND-CRAFTED
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HAND-CRAFTEDPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 52.14 | 3.11 | |
| ELIGN_advPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 19.46 | 5.05 | |
| ELIGN_selfPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 12.84 | 4.29 | |
| CURIO_selfPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 11.88 | 2.88 | |
| CURIO_teamPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 9.54 | 5.04 | |
| SPARSEPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 4.58 | 3.27 | |
| ELIGN_teamPartial observability=true, Decentralized training=true, Agt# v Adv#=2v22022.10 | 2.96 | 4.03 |