Dataset Cards - OMAR
simple_spread - Download
Metadata
| Environment name | Version | Agents | Action type | Observation size | Reward type |
|---|---|---|---|---|---|
| MPE | Code included in OMAR repository | 3 | Discrete | [18] | Dense |
Generation procedure for each dataset
Converted from omar format to a Vault.
Summary statistics
| Uid | Episode return mean | Min return | Max return | Transitions | Trajectories | Joint SACo |
|---|---|---|---|---|---|---|
| Random | 159.57 ± 60.46 | -5.43 | 510.05 | 1000000 | 40000 | 1.00 |
| Medium-Replay | 203.74 ± 80.49 | 35.69 | 582.09 | 97500 | 3900 | 1.00 |
| Medium | 273.39 ± 92.06 | 27.35 | 649.51 | 1000000 | 40000 | 1.00 |
| Expert | 530.95 ± 71.41 | 54.96 | 743.89 | 1000000 | 40000 | 1.00 |
simple_tag - Download
Metadata
| Environment name | Version | Agents | Action type | Observation size | Reward type |
|---|---|---|---|---|---|
| MPE | Code included in OMAR repository | 4 | Discrete | [16] | Dense |
Generation procedure for each dataset
Converted from omar format to a Vault.
Summary statistics
| Uid | Episode return mean | Min return | Max return | Transitions | Trajectories | Joint SACo |
|---|---|---|---|---|---|---|
| Random | -4.13 ± 10.81 | -20.18 | 117.09 | 1000000 | 40000 | 1.00 |
| Medium-Replay | 3.90 ± 20.28 | -17.11 | 146.12 | 62500 | 2500 | 1.00 |
| Medium | 116.36 ± 58.86 | -12.66 | 418.25 | 1000000 | 40000 | 1.00 |
| Expert | 207.90 ± 77.51 | -16.04 | 549.20 | 1000000 | 40000 | 1.00 |
simple_world - Download
Metadata
| Environment name | Version | Agents | Action type | Observation size | Reward type |
|---|---|---|---|---|---|
| MPE | Code included in OMAR repository | 4 | Discrete | [24] | Dense |
Generation procedure for each dataset
Converted from omar format to a Vault.
Summary statistics
| Uid | Episode return mean | Min return | Max return | Transitions | Trajectories | Joint SACo |
|---|---|---|---|---|---|---|
| Random | -6.83 ± 5.74 | -17.81 | 54.41 | 1000000 | 40000 | 1.00 |
| Medium-Replay | 1.23 ± 13.49 | -17.56 | 112.90 | 80000 | 3200 | 1.00 |
| Medium | 65.86 ± 29.55 | -9.15 | 198.82 | 1000000 | 40000 | 1.00 |
| Expert | 85.21 ± 31.11 | -11.55 | 238.70 | 1000000 | 40000 | 1.00 |
2halfcheetah - Download
Metadata
| Environment name | Version | Agents | Action type | Observation size | Reward type |
|---|---|---|---|---|---|
| MAMuJoCo | V1.0, Mujoco v200 | 2 | Continuous | [6] | Dense |
Generation procedure for each dataset
Converted from omar format to a Vault.
Summary statistics
| Uid | Episode return mean | Min return | Max return | Transitions | Trajectories | Joint SACo |
|---|---|---|---|---|---|---|
| Random | -282.89 ± 77.50 | -516.90 | -62.62 | 1000000 | 1000 | 1.00 |
| Medium-Replay | 423.49 ± 655.68 | -509.10 | 1993.00 | 460000 | 460 | 1.00 |
| Medium | 1568.87 ± 273.38 | 20.49 | 1904.56 | 1000000 | 1000 | 1.00 |
| Expert | 3338.69 ± 252.58 | 852.45 | 3605.42 | 1000000 | 1000 | 1.00 |