mirror of
https://github.com/saymrwulf/stable-baselines3.git
synced 2026-11-16 08:49:07 +00:00
Fix VectorizedActionNoise in OffPolicyAlgorithm (#1657)
* moves VectorizedActionNoise into _setup_learn() * update changelog --------- Co-authored-by: Antonin RAFFIN <antonin.raffin@ensta.org> Co-authored-by: Antonin Raffin <antonin.raffin@dlr.de>
This commit is contained in:
parent
5c93e9f426
commit
c99d65c664
2 changed files with 10 additions and 5 deletions
|
|
@ -20,7 +20,8 @@ New Features:
|
||||||
|
|
||||||
Bug Fixes:
|
Bug Fixes:
|
||||||
^^^^^^^^^^
|
^^^^^^^^^^
|
||||||
- Prevents OOB error on Windows if no seed is passed (@PatrickHelm)
|
- Moves VectorizedActionNoise into ``_setup_learn()`` in OffPolicyAlgorithm (@PatrickHelm)
|
||||||
|
- Prevents out of bound error on Windows if no seed is passed (@PatrickHelm)
|
||||||
|
|
||||||
Deprecations:
|
Deprecations:
|
||||||
^^^^^^^^^^^^^
|
^^^^^^^^^^^^^
|
||||||
|
|
|
||||||
|
|
@ -281,6 +281,14 @@ class OffPolicyAlgorithm(BaseAlgorithm):
|
||||||
pos = (replay_buffer.pos - 1) % replay_buffer.buffer_size
|
pos = (replay_buffer.pos - 1) % replay_buffer.buffer_size
|
||||||
replay_buffer.dones[pos] = True
|
replay_buffer.dones[pos] = True
|
||||||
|
|
||||||
|
# Vectorize action noise if needed
|
||||||
|
if (
|
||||||
|
self.action_noise is not None
|
||||||
|
and self.env.num_envs > 1
|
||||||
|
and not isinstance(self.action_noise, VectorizedActionNoise)
|
||||||
|
):
|
||||||
|
self.action_noise = VectorizedActionNoise(self.action_noise, self.env.num_envs)
|
||||||
|
|
||||||
return super()._setup_learn(
|
return super()._setup_learn(
|
||||||
total_timesteps,
|
total_timesteps,
|
||||||
callback,
|
callback,
|
||||||
|
|
@ -523,10 +531,6 @@ class OffPolicyAlgorithm(BaseAlgorithm):
|
||||||
if env.num_envs > 1:
|
if env.num_envs > 1:
|
||||||
assert train_freq.unit == TrainFrequencyUnit.STEP, "You must use only one env when doing episodic training."
|
assert train_freq.unit == TrainFrequencyUnit.STEP, "You must use only one env when doing episodic training."
|
||||||
|
|
||||||
# Vectorize action noise if needed
|
|
||||||
if action_noise is not None and env.num_envs > 1 and not isinstance(action_noise, VectorizedActionNoise):
|
|
||||||
action_noise = VectorizedActionNoise(action_noise, env.num_envs)
|
|
||||||
|
|
||||||
if self.use_sde:
|
if self.use_sde:
|
||||||
self.actor.reset_noise(env.num_envs)
|
self.actor.reset_noise(env.num_envs)
|
||||||
|
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue