Skip to content

Commit 83ca375

Browse files
author
Morvan Zhou
committed
update
1 parent b90547f commit 83ca375

File tree

1 file changed

+7
-6
lines changed

1 file changed

+7
-6
lines changed

contents/5_Deep_Q_Network/DQN_modified.py

Lines changed: 7 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -53,13 +53,14 @@ def __init__(
5353

5454
t_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='target_net')
5555
e_params = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='eval_net')
56-
self.target_replace_op = [tf.assign(t, e) for t, e in zip(t_params, e_params)]
56+
57+
with tf.variable_scope('soft_replacement'):
58+
self.target_replace_op = [tf.assign(t, e) for t, e in zip(t_params, e_params)]
5759

5860
self.sess = tf.Session()
5961

6062
if output_graph:
6163
# $ tensorboard --logdir=logs
62-
# tf.train.SummaryWriter soon be deprecated, use following
6364
tf.summary.FileWriter("logs/", self.sess.graph)
6465

6566
self.sess.run(tf.global_variables_initializer())
@@ -77,16 +78,16 @@ def _build_net(self):
7778
# ------------------ build evaluate_net ------------------
7879
with tf.variable_scope('eval_net'):
7980
e1 = tf.layers.dense(self.s, 20, tf.nn.relu, kernel_initializer=w_initializer,
80-
bias_initializer=b_initializer)
81+
bias_initializer=b_initializer, name='e1')
8182
self.q_eval = tf.layers.dense(e1, self.n_actions, kernel_initializer=w_initializer,
82-
bias_initializer=b_initializer)
83+
bias_initializer=b_initializer, name='q')
8384

8485
# ------------------ build target_net ------------------
8586
with tf.variable_scope('target_net'):
8687
t1 = tf.layers.dense(self.s_, 20, tf.nn.relu, kernel_initializer=w_initializer,
87-
bias_initializer=b_initializer)
88+
bias_initializer=b_initializer, name='t1')
8889
self.q_next = tf.layers.dense(t1, self.n_actions, kernel_initializer=w_initializer,
89-
bias_initializer=b_initializer)
90+
bias_initializer=b_initializer, name='t2')
9091

9192
with tf.variable_scope('q_target'):
9293
q_target = self.r + self.gamma * tf.reduce_max(self.q_next, axis=1, name='Qmax_s_') # shape=(None, )

0 commit comments

Comments
 (0)