Fix missing learning rate and entropy coeff schedule for torch PPO (#8572)

This commit is contained in:
Jan Blumenkamp
2020-05-23 10:54:18 -07:00
committed by GitHub
parent 2ab1b773d4
commit d6f78f58dc
+4 -1
View File
@@ -223,4 +223,7 @@ PPOTorchPolicy = build_torch_policy(
extra_grad_process_fn=apply_grad_clipping,
before_init=setup_config,
after_init=setup_mixins,
mixins=[KLCoeffMixin, ValueNetworkMixin])
mixins=[
LearningRateSchedule, EntropyCoeffSchedule, KLCoeffMixin,
ValueNetworkMixin
])