Use ssf also for pca
This commit is contained in:
+1
-1
@@ -208,7 +208,7 @@ class PPO(BetterOnPolicyAlgorithm):
|
||||
actions = rollout_data.actions.long().flatten()
|
||||
|
||||
# Re-sample the noise matrix because the log_std has changed
|
||||
if self.use_sde:
|
||||
if self.use_sde or self.use_pca:
|
||||
self.policy.reset_noise(self.batch_size)
|
||||
|
||||
values, log_prob, entropy = self.policy.evaluate_actions(rollout_data.observations, actions)
|
||||
|
||||
Reference in New Issue
Block a user