ae summary

autoencoder.py

@@ -50,9 +50,9 @@ def trainPixelCNNAE(conf, data):
     decoder_X = tf.placeholder(tf.float32, shape=[None, conf.img_height, conf.img_width, conf.channel])
 
     encoder = AE(encoder_X)
-    conf.num_classes = int(encoder.pred.get_shape()[1])
     decoder = PixelCNN(decoder_X, conf, encoder.pred)
     y = decoder.pred
+    tf.scalar_summary('loss', decoder.loss)
 
     #loss = tf.reduce_mean(tf.square(encoder_X - y))
     #trainer = tf.train.GradientDescentOptimizer(1e-4).minimize(loss)
@@ -65,24 +65,34 @@ def trainPixelCNNAE(conf, data):
 
     saver = tf.train.Saver(tf.trainable_variables())
     with tf.Session() as sess:
+        merged = tf.merge_all_summaries()
+        writer = tf.train.SummaryWriter('/tmp/mnist_ae', sess.graph)
+
         sess.run(tf.initialize_all_variables())
+
         if os.path.exists(conf.ckpt_file):
             saver.restore(sess, conf.ckpt_file)
             print "Model Restored"
 
         pointer = 0
+        step = 0
         for i in range(conf.epochs):
             for j in range(conf.num_batches):
                 if conf.data == 'mnist':
                     batch_X = binarize(data.train.next_batch(conf.batch_size)[0].reshape(conf.batch_size, conf.img_height, conf.img_width, conf.channel))
                 else:
                     batch_X, pointer = get_batch(data, pointer, conf.batch_size)
 
-                _, l = sess.run([optimizer, decoder.loss], feed_dict={encoder_X: batch_X, decoder_X: batch_X})
+                _, l, summary = sess.run([optimizer, decoder.loss, merged], feed_dict={encoder_X: batch_X, decoder_X: batch_X})
+                writer.add_summary(summary, step)
+                step += 1
             print "Epoch: %d, Cost: %f"%(i, l)
+            if (i+1)%10 == 0:
+                saver.save(sess, conf.ckpt_file)
+                generate_ae(sess, encoder_X, decoder_X, y, data, conf, str(i))
 
-        saver.save(sess, conf.ckpt_file)
-        #generate_ae(sess, encoder_X, decoder_X, y, data, conf)
+        writer.close()
+        '''
         data = input_data.read_data_sets("data/")
         n_examples = 10
         if conf.data == 'mnist':
@@ -97,7 +107,7 @@ def trainPixelCNNAE(conf, data):
         fig.show()
         plt.draw()
         plt.waitforbuttonpress()
-
+        '''
 
 
 if __name__ == "__main__":
@@ -113,8 +123,10 @@ class Conf(object):
     conf.layers=5
     conf.samples_path='samples/ae'
     conf.ckpt_path='ckpts/ae'
-    conf.epochs=10
-    conf.batch_size = 100
+    conf.summary_path='/tmp/mnist_ae'
+    conf.epochs=50
+    conf.batch_size = 64
+    conf.num_classes = 10
 
     if conf.data == 'mnist':
         from tensorflow.examples.tutorials.mnist import input_data
@@ -136,5 +148,9 @@ class Conf(object):
 
     conf.num_batches = train_size // conf.batch_size
     conf = makepaths(conf)
+    if tf.gfile.Exists(conf.summary_path):
+        tf.gfile.DeleteRecursively(conf.summary_path)
+    tf.gfile.MakeDirs(conf.summary_path)
+
     trainPixelCNNAE(conf, data)