Update DDPG

2026-09-11 11:53:01 +08:00 · 2018-01-19 11:24:32 -07:00
parent db49a76c53
commit eb16eeb2a1
3 changed files with 20 additions and 29 deletions
@@ -97,7 +97,7 @@ class DDPGAgent:
                actions = actor.predict(states, False)
                var_actions = Variable(actions.data, requires_grad=True)
                q = critic.predict(states, var_actions)
-                q.backward(torch.ones(q.size()))
+                q.backward(critic.FloatTensor(np.ones(q.size())))

                actor.zero_grad()
                self.actor_opt.zero_grad()