update a little

icoxfog417 · Nov 18, 2018 · be7148d · be7148d
1 parent 331663b
commit be7148d
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 2 deletions.
diff --git a/FN/a2c_agent.py b/FN/a2c_agent.py
@@ -66,11 +66,12 @@ def set_updater(self, optimizer,
 
         _, action_evals, values = self.model.output
 
-        advantages = rewards - values
         neg_logs = tf.nn.sparse_softmax_cross_entropy_with_logits(
                         logits=action_evals, labels=actions)
+        advantages = rewards - values
 
         policy_loss = tf.reduce_mean(neg_logs * tf.nn.softplus(advantages))
+
         batch_indices = tf.range(tf.shape(actions)[0])
         action_indices = tf.stack([batch_indices, actions], axis=1)
         value_only = rewards - tf.gather_nd(action_evals, action_indices)

diff --git a/IM/dagger.py b/IM/dagger.py
@@ -127,7 +127,7 @@ def imitate(self, env, teacher, initial_step=100, train_step=200,
         states = []
         actions = []
 
-        # Collect teacher's demonstration.
+        # Collect teacher's demonstrations.
         for e in range(initial_step):
             s = env.reset()
             done = False