def testConstructWrapperWithExistingEmptyDumpRoot(self):
os.mkdir(self._tmp_dir)
self.assertTrue(os.path.isdir(self._tmp_dir))
local_cli.LocalCLIDebugWrapperSession(session.Session(),
dump_root=self._tmp_dir,
log_usage=False)
def testConstructWrapperWithExistingFileDumpRoot(self):
os.mkdir(self._tmp_dir)
file_path = os.path.join(self._tmp_dir, "foo")
open(file_path, "a").close() # Create the file
self.assertTrue(os.path.isfile(file_path))
with self.assertRaisesRegexp(ValueError,
"dump_root path points to a file"):
local_cli.LocalCLIDebugWrapperSession(session.Session(),
dump_root=file_path)
def testConstructWrapperWithExistingNonEmptyDumpRoot(self):
os.mkdir(self._tmp_dir)
dir_path = os.path.join(self._tmp_dir, "foo")
os.mkdir(dir_path)
self.assertTrue(os.path.isdir(dir_path))
with self.assertRaisesRegexp(
ValueError, "dump_root path points to a non-empty directory"):
local_cli.LocalCLIDebugWrapperSession(session.Session(),
dump_root=self._tmp_dir)
def main(_):
sess = tf.Session()
# Construct the TensorFlow network.
n0 = tf.Variable(np.ones([FLAGS.tensor_size] * 2), name="node_00")
n1 = tf.Variable(np.ones([FLAGS.tensor_size] * 2), name="node_01")
if FLAGS.length > 100:
raise ValueError("n is too big.")
for i in xrange(2, FLAGS.length):
n0, n1 = n1, tf.add(n0, n1, name="node_%.2d" % i)
sess.run(tf.initialize_all_variables())
# Wrap the TensorFlow Session object for debugging.
sess = local_cli.LocalCLIDebugWrapperSession(sess)
sess.run(n1)
def testConstructWrapper(self):
local_cli.LocalCLIDebugWrapperSession(session.Session(),
log_usage=False)
def main(_):
# Import data
mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)
def feed_dict(train):
if train:
xs, ys = mnist.train.next_batch(FLAGS.train_batch_size, fake_data=False)
else:
xs, ys = mnist.test.images, mnist.test.labels
return {x: xs, y_: ys}
sess = tf.InteractiveSession()
# Create the MNIST neural network graph.
# Input placeholders.
with tf.name_scope("input"):
x = tf.placeholder(
tf.float32, [None, IMAGE_SIZE * IMAGE_SIZE], name="x-input")
y_ = tf.placeholder(tf.float32, [None, NUM_LABELS], name="y-input")
def weight_variable(shape):
"""Create a weight variable with appropriate initialization."""
initial = tf.truncated_normal(shape, stddev=0.1, seed=RAND_SEED)
return tf.Variable(initial)
def bias_variable(shape):
"""Create a bias variable with appropriate initialization."""
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
def nn_layer(input_tensor, input_dim, output_dim, layer_name, act=tf.nn.relu):
"""Reusable code for making a simple neural net layer."""
# Adding a name scope ensures logical grouping of the layers in the graph.
with tf.name_scope(layer_name):
# This Variable will hold the state of the weights for the layer
with tf.name_scope("weights"):
weights = weight_variable([input_dim, output_dim])
with tf.name_scope("biases"):
biases = bias_variable([output_dim])
with tf.name_scope("Wx_plus_b"):
preactivate = tf.matmul(input_tensor, weights) + biases
activations = act(preactivate)
return activations
hidden = nn_layer(x, IMAGE_SIZE**2, HIDDEN_SIZE, "hidden")
y = nn_layer(hidden, HIDDEN_SIZE, NUM_LABELS, "softmax", act=tf.nn.softmax)
with tf.name_scope("cross_entropy"):
# The following line is the culprit of the bad numerical values that appear
# during training of this graph. Log of zero gives inf, which is first seen
# in the intermediate tensor "cross_entropy/Log:0" during the 4th run()
# call. A multiplication of the inf values with zeros leads to nans,
# which is first in "cross_entropy/mul:0".
#
# You can use clipping to fix this issue, e.g.,
# diff = y_ * tf.log(tf.clip_by_value(y, 1e-8, 1.0))
diff = y_ * tf.log(y)
with tf.name_scope("total"):
cross_entropy = -tf.reduce_mean(diff)
with tf.name_scope("train"):
train_step = tf.train.AdamOptimizer(FLAGS.learning_rate).minimize(
cross_entropy)
with tf.name_scope("accuracy"):
with tf.name_scope("correct_prediction"):
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
with tf.name_scope("accuracy"):
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
sess.run(tf.initialize_all_variables())
if FLAGS.debug:
sess = local_cli.LocalCLIDebugWrapperSession(sess)
sess.add_tensor_filter("has_inf_or_nan", debug_data.has_inf_or_nan)
# Add this point, sess is a debug wrapper around the actual Session if
# FLAGS.debug is true. In that case, calling run() will launch the CLI.
for i in range(FLAGS.max_steps):
acc = sess.run(accuracy, feed_dict=feed_dict(False))
print("Accuracy at step %d: %s" % (i, acc))
sess.run(train_step, feed_dict=feed_dict(True))
def testConstructWrapper(self):
local_cli.LocalCLIDebugWrapperSession(session.Session())