Finalized factoring out projections
This commit is contained in:
@@ -336,8 +336,7 @@ class PPO(GaussianRolloutCollectorAuxclass, OnPolicyAlgorithm):
|
||||
|
||||
policy_loss = surrogate_loss + self.ent_coef * entropy_loss + \
|
||||
trust_region_loss + self.action_coef * action_loss
|
||||
import pdb
|
||||
pdb.set_trace()
|
||||
|
||||
pg_losses.append(policy_loss.item())
|
||||
|
||||
loss = policy_loss + self.vf_coef * value_loss
|
||||
|
||||
Reference in New Issue
Block a user