OscervaME
/
stable-baselines3-contrib-sacd
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Rename QRDQN logger key (#67)

This commit is contained in:
Antonin RAFFIN 2022-04-12 12:50:35 +02:00 committed by GitHub
parent 99853265a9
commit 812648e6cd
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
docs/misc/changelog.rst
View File

@ -11,6 +11,7 @@ Breaking Changes:
- Upgraded to Stable-Baselines3 >= 1.5.1a1 - Upgraded to Stable-Baselines3 >= 1.5.1a1
- Changed the way policy "aliases" are handled ("MlpPolicy", "CnnPolicy", ...), removing the former - Changed the way policy "aliases" are handled ("MlpPolicy", "CnnPolicy", ...), removing the former
``register_policy`` helper, ``policy_base`` parameter and using ``policy_aliases`` static attributes instead (@Gregwar) ``register_policy`` helper, ``policy_base`` parameter and using ``policy_aliases`` static attributes instead (@Gregwar)
- Renamed ``rollout/exploration rate`` key to ``rollout/exploration_rate`` for QRDQN (to be consistent with SB3 DQN)
New Features: New Features:
^^^^^^^^^^^^^ ^^^^^^^^^^^^^

2
sb3_contrib/qrdqn/qrdqn.py
View File

@ -159,7 +159,7 @@ class QRDQN(OffPolicyAlgorithm):
polyak_update(self.quantile_net.parameters(), self.quantile_net_target.parameters(), self.tau) polyak_update(self.quantile_net.parameters(), self.quantile_net_target.parameters(), self.tau)
self.exploration_rate = self.exploration_schedule(self._current_progress_remaining) self.exploration_rate = self.exploration_schedule(self._current_progress_remaining)
self.logger.record("rollout/exploration rate", self.exploration_rate) self.logger.record("rollout/exploration_rate", self.exploration_rate)
def train(self, gradient_steps: int, batch_size: int = 100) -> None: def train(self, gradient_steps: int, batch_size: int = 100) -> None:
# Switch to train mode (this affects batch norm / dropout) # Switch to train mode (this affects batch norm / dropout)