Updated DQN optimizer input to only include q_network parameters as input (#1963)

* Updated DQN optimizer input to only include q_network parameters

* Update version

---------

Co-authored-by: Antonin RAFFIN <antonin.raffin@ensta.org>
This commit is contained in:
Corentin 2024-07-05 19:07:55 +02:00 committed by GitHub
parent 0eebde7ca1
commit d8148deeaa
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 4 additions and 3 deletions

View file

@ -3,7 +3,7 @@
Changelog Changelog
========== ==========
Release 2.4.0a4 (WIP) Release 2.4.0a5 (WIP)
-------------------------- --------------------------
Breaking Changes: Breaking Changes:
@ -20,6 +20,7 @@ Bug Fixes:
- ``CallbackList`` now sets the ``.parent`` attribute of child callbacks to its own ``.parent``. (will-maclean) - ``CallbackList`` now sets the ``.parent`` attribute of child callbacks to its own ``.parent``. (will-maclean)
- Fixed error when loading a model that has ``net_arch`` manually set to ``None`` (@jak3122) - Fixed error when loading a model that has ``net_arch`` manually set to ``None`` (@jak3122)
- Set requirement numpy<2.0 until PyTorch is compatible (https://github.com/pytorch/pytorch/issues/107302) - Set requirement numpy<2.0 until PyTorch is compatible (https://github.com/pytorch/pytorch/issues/107302)
- Updated DQN optimizer input to only include q_network parameters, removing the target_q_network ones (@corentinlger)
`SB3-Contrib`_ `SB3-Contrib`_
^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^

View file

@ -167,7 +167,7 @@ class DQNPolicy(BasePolicy):
# Setup optimizer with initial learning rate # Setup optimizer with initial learning rate
self.optimizer = self.optimizer_class( # type: ignore[call-arg] self.optimizer = self.optimizer_class( # type: ignore[call-arg]
self.parameters(), self.q_net.parameters(),
lr=lr_schedule(1), lr=lr_schedule(1),
**self.optimizer_kwargs, **self.optimizer_kwargs,
) )

View file

@ -1 +1 @@
2.4.0a4 2.4.0a5