A HRL framework based on stable-baseline3.
| Features | HRL:goal-conditioned hrl |
|---|---|
| High: on-policy Low: on-policy | ✔️ |
| High: on-policy Low: off-policy | ✔️ |
| High: off-policy Low: off-policy | ✔️ |
| High: off-policy Low: on-policy | ❌ |
| Tensorboard support | ✔️ |
| Learned goal space | ❌ |
| Hindsight Experience replay | ✔️ |
| Goal correction | ✔️ |
| 3-levels or more | ❌ |
| dynamic low steps | ❌ |