def train_epoch(flag=False, initial=True):
	diter = 5
	count =  0
	large_iter =  100
	if flag  :
		final_iter = large_iter
	else:
		final_iter = diter
	run=0
	start_time = time.time()
	loss_val = [0,0]
	while run <= num_examples:
		for t in range(final_iter):
			feed_list = gan.generate_batch()
			run += batch_size
			feed_dict = {
				placeholders['image_input'] : feed_list[0],
				placeholders['x'] : feed_list[1],
				placeholders['image_class_input'] : feed_list[2],
			}
			_, loss_val[0] = session.run([optimizers["discriminator"],losses["disc_image_discriminator"]], feed_dict=feed_dict)

		for _ in range(2*diter):
			feed_list = gan.generate_batch()
			run += batch_size
			feed_dict = {
				placeholders['image_input'] : feed_list[0],
				placeholders['image_class_input'] : feed_list[2],
			}
			if initial :
				_, loss_val[6] = session.run([optimizers["generator"], losses["generator_image"]], feed_dict=feed_dict)
			else:
				_, loss_val[6] = session.run([optimizers["generator_gan"], losses["generator_image"]], feed_dict=feed_dict)

		# z_c = session.run(z_hat_c, feed_dict=feed_dict)
		count += 1
		if count % 10 == 0 or flag:
			print("%d:%d : "%(ep+1,run) + " : ".join(map(lambda x : str(x),loss_val)) + " " + str(time.time() - start_time))
			# print(z_c)
		start_time = time.time()