|
@@ -113,7 +113,7 @@ square the agent must get to. This environment is extremely difficult to
|
|
|
solve using classical RL. However, by gradually increasing the number of
|
|
solve using classical RL. However, by gradually increasing the number of
|
|
|
rooms and building a curriculum, the environment can be solved.
|
|
rooms and building a curriculum, the environment can be solved.
|
|
|
|
|
|
|
|
-### Fetch Environment
|
|
|
|
|
|
|
+### Fetch environment
|
|
|
|
|
|
|
|
Registered configurations:
|
|
Registered configurations:
|
|
|
- `MiniGrid-Fetch-8x8-v0`
|
|
- `MiniGrid-Fetch-8x8-v0`
|
|
@@ -127,7 +127,7 @@ agent receives a textual string as part of its observation telling it
|
|
|
which object to pick up. Picking up the wrong object produces a negative
|
|
which object to pick up. Picking up the wrong object produces a negative
|
|
|
reward.
|
|
reward.
|
|
|
|
|
|
|
|
-### Four Room Question Answering
|
|
|
|
|
|
|
+### Four room question answering environment
|
|
|
|
|
|
|
|
Registered configurations:
|
|
Registered configurations:
|
|
|
- `MiniGrid-FourRoomQA-v0`
|
|
- `MiniGrid-FourRoomQA-v0`
|
Maxime Chevalier-Boisvert