Dominik Jain ca69e79b4a Change the way in which deterministic evaluation is controlled: ...

* Remove flag `eval_mode` from Collector.collect
  * Replace flag `is_eval` in BasePolicy with `is_within_training_step` (negating usages)
    and set it appropriately in BaseTrainer

2024-05-03 15:18:39 +02:00

..

atari

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

box2d

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

discrete

Fix invalid kwarg

2024-05-03 10:12:41 +02:00

inverse

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

modelbased

Fix SAC loss explode (#333)

2021-04-04 17:33:35 +08:00

mujoco

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

offline

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

vizdoom

Change the way in which deterministic evaluation is controlled:

2024-05-03 15:18:39 +02:00

__init__.py

Fix critic network for Discrete CRR (#485)

2021-11-28 23:10:28 +08:00