gryf/coach
1
0 Fork 0
mirror of https://github.com/gryf/coach.git synced 2026-10-10 07:52:33 +02:00
Code Issues Projects Releases Wiki Activity

fix bug in ddpg

Browse Source
This commit is contained in:
Zach Dwiel committed 2018-02-21 10:05:57 -05:00
1 parent 8248caf35e
commit 5cf10e5f52
2 files changed
+2 -3
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

agents/ddpg_agent.py
+1 -1
View File
@@ -54,7 +54,7 @@ class DDPGAgent(ActorCriticAgent):
actions_mean = self.actor_network.online_network.predict(current_states)
critic_online_network = self.critic_network.online_network
# TODO: convert into call to predict, current method ignores lstm middleware for example
action_gradients = self.critic_network.sess.run(critic_online_network.gradients_wrt_inputs[1],
action_gradients = self.critic_network.sess.run(critic_online_network.gradients_wrt_inputs['action'],
feed_dict=critic_online_network._feed_dict({
**current_states,
'action': actions_mean,