From d6f78f58dc7548217b46565a3dd42cd2e0133e66 Mon Sep 17 00:00:00 2001
From: Jan Blumenkamp <jb2270@cam.ac.uk>
Date: Sat, 23 May 2020 19:54:18 +0200
Subject: [PATCH] Fix missing learning rate and entropy coeff schedule for
 torch PPO (#8572)

---
 rllib/agents/ppo/ppo_torch_policy.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/rllib/agents/ppo/ppo_torch_policy.py b/rllib/agents/ppo/ppo_torch_policy.py
index 7229b251e..a878b70d8 100644
--- a/rllib/agents/ppo/ppo_torch_policy.py
+++ b/rllib/agents/ppo/ppo_torch_policy.py
@@ -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
+    ])