with tf.variable_scope('optimizer'): d_optim = tf.train.AdamOptimizer(learning_rate= 1e-4 ).minimize(train_d_loss, var_list=d_param) g_optim = tf.train.AdamOptimizer(learning_rate= 1e-4 ).minimize(train_g_loss, var_list=g_param) saver = tf.train.Saver() config = tf.ConfigProto() config.gpu_options.allow_growth = True session = tf.Session(config=config) with tf.Session() as sess: if Is_train == False:# for testing the trained model saver.restore(sess,'./GANmodel.ckpt') writer = tf.summary.FileWriter('./graphs', sess.graph) if Is_train == True:# for training the model training_data = data_loader_special.load_data_wrapper() #uploading the data tvals = np.repeat(np.linspace(0.1,1.9,n_temps),10000) c = list(zip(training_data,tvals)) random.shuffle(c) # pairing and shuffling the data and temeperature training_data, tvals = zip(*c) print(len(training_data),len(tvals)) m = tf.placeholder(tf.float32,[datapoints, lattice_size, lattice_size,1]) n = tf.placeholder(tf.float32,[datapoints,lattice_size+2,lattice_size+2,1]) b = tf.placeholder(tf.float32,[datapoints, n_z]) # Uploading the data to prevent memeory issues prefetch and batching dataset = tf.data.Dataset.from_tensor_slices((m,n,b)) dataset = dataset.prefetch(buffer_size=100) dataset = dataset.batch(batch_size) iterator = dataset.make_initializable_iterator() next = iterator.get_next()
# 4. Update weights g_param = tf.trainable_variables(scope='Generator') d_param = tf.trainable_variables(scope='Discriminator') print(tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES)) with tf.name_scope('optimizer'): d_optim = tf.train.AdamOptimizer(learning_rate=1e-4).minimize( train_d_loss, var_list=d_param) g_optim = tf.train.AdamOptimizer(learning_rate=1e-3).minimize( train_g_loss, var_list=g_param) saver = tf.train.Saver() with tf.Session() as sess: # saver.restore(sess,'./GANmodel.ckpt') # writer = tf.summary.FileWriter('./graphs', sess.graph) training_data = data_loader_special.load_data_wrapper() tvals = np.repeat(np.linspace(0.1, 2.0, 32), 10000) c = list(zip(training_data, tvals)) random.shuffle(c) training_data, tvals = zip(*c) print(len(training_data), len(tvals)) m = tf.placeholder(tf.float32, [datapoints, 128]) n = tf.placeholder(tf.float32, [datapoints, 1]) dataset = tf.data.Dataset.from_tensor_slices((m, n)) dataset = dataset.prefetch(buffer_size=1000) dataset = dataset.batch(batch_size) iterator = dataset.make_initializable_iterator() next = iterator.get_next() print("============< WARNING >===============") sess.run(tf.global_variables_initializer()) print("==========< Model DELETED >===========")