tf_mnist2.py

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import math
import os
import sys
import time

import tensorflow as tf
from tensorflow.contrib.learn.python.learn.datasets.mnist import read_data_sets

NUM_CLASSES = 10
IMAGE_SIZE = 28
IMAGE_PIXELS = IMAGE_SIZE * IMAGE_SIZE

def inference(images, hidden1_units, hidden2_units):
  # Hidden 1
  with tf.name_scope('hidden1'):
    weights = tf.Variable(
        tf.truncated_normal([IMAGE_PIXELS, hidden1_units],
                            stddev=1.0 / math.sqrt(float(IMAGE_PIXELS))),
        name='weights')
    biases = tf.Variable(tf.zeros([hidden1_units]),
                         name='biases')
    hidden1 = tf.nn.relu(tf.matmul(images, weights) + biases)
  # Hidden 2
  with tf.name_scope('hidden2'):
    weights = tf.Variable(
        tf.truncated_normal([hidden1_units, hidden2_units],
                            stddev=1.0 / math.sqrt(float(hidden1_units))),
        name='weights')
    biases = tf.Variable(tf.zeros([hidden2_units]),
                         name='biases')
    hidden2 = tf.nn.relu(tf.matmul(hidden1, weights) + biases)
  # Linear
  with tf.name_scope('softmax_linear'):
    weights = tf.Variable(
        tf.truncated_normal([hidden2_units, NUM_CLASSES],
                            stddev=1.0 / math.sqrt(float(hidden2_units))),
        name='weights')
    biases = tf.Variable(tf.zeros([NUM_CLASSES]),
                         name='biases')
    logits = tf.matmul(hidden2, weights) + biases
  return logits


def loss(logits, labels):
  labels = tf.to_int64(labels)
  return tf.losses.sparse_softmax_cross_entropy(labels=labels, logits=logits)


def training(loss, learning_rate, name='train_op'):
  # tf.summary.scalar('loss', loss)
  optimizer = tf.train.GradientDescentOptimizer(learning_rate)
  global_step = tf.Variable(0, name='global_step', trainable=False)
  train_op = optimizer.minimize(loss, global_step=global_step, name=name)
  return train_op


def evaluation(logits, labels, name='eval_correct'):
  correct = tf.nn.in_top_k(logits, labels, 1)
  return tf.reduce_sum(tf.cast(correct, tf.int32), name=name)


input_data_dir = os.path.join(os.getenv('TEST_TMPDIR', '/tmp'), 'tensorflow/mnist/input_data')
data_sets = read_data_sets(input_data_dir, False)
log_dir = os.path.join(os.getenv('TEST_TMPDIR', '/tmp'), 'tensorflow/mnist/logs/fully_connected_feed'),
batch_size = 100
learning_rate = 0.01
hidden1 = 128
hidden2 = 32
max_steps = 2000
checkpoint_file = os.path.join(os.getenv('HOME'), 'Tmp/tf_converter/model2.ckpt')
meta_file = checkpoint_file + '.meta'

### Utils

def placeholder_inputs(batch_size):
  images_placeholder = tf.placeholder(
    tf.float32, shape=(batch_size, IMAGE_PIXELS), name='images_placeholder')
  labels_placeholder = tf.placeholder(tf.int32, shape=(batch_size), name='labels_placeholder')
  return images_placeholder, labels_placeholder

def fill_feed_dict(data_set, images_pl, labels_pl):
  images_feed, labels_feed = data_set.next_batch(batch_size, False)
  feed_dict = {
      images_pl: images_feed,
      labels_pl: labels_feed,
  }
  return feed_dict

def do_eval(sess,
            eval_correct,
            images_placeholder,
            labels_placeholder,
            data_set):
  # And run one epoch of eval.
  true_count = 0  # Counts the number of correct predictions.
  steps_per_epoch = data_set.num_examples // batch_size
  num_examples = steps_per_epoch * batch_size

  for step in xrange(steps_per_epoch):
    feed_dict = fill_feed_dict(data_set,
                               images_placeholder,
                               labels_placeholder)
    true_count += sess.run(eval_correct, feed_dict=feed_dict)
  precision = float(true_count) / num_examples
  print('Num examples: %d  Num correct: %d  Precision @ 1: %0.04f' %
        (num_examples, true_count, precision))


## Build Graph and save

with tf.Graph().as_default():
  sess = tf.Session()
  images_placeholder, labels_placeholder = placeholder_inputs(batch_size)
  logits = inference(images_placeholder, 128, 32)
  loss = loss(logits, labels_placeholder)
  train_op = training(loss, learning_rate)
  eval_correct = evaluation(logits, labels_placeholder)
  saver = tf.train.Saver()

  init = tf.global_variables_initializer()
  sess.run(init)

  saver.save(sess, checkpoint_file)


## Start training

with tf.Graph().as_default():
  sess = tf.Session()
  saver = tf.train.import_meta_graph(meta_file)
  graph = tf.get_default_graph()

  # Not sure about this part though...
  init = tf.global_variables_initializer()
  sess.run(init)

  images_placeholder = graph.get_tensor_by_name('images_placeholder:0')
  labels_placeholder = graph.get_tensor_by_name('labels_placeholder:0')
  eval_correct = graph.get_tensor_by_name('eval_correct:0')

  train_op = tf.get_collection('train_op')[0]
  loss = tf.get_collection('losses')[0]

  for step in xrange(max_steps):
    start_time = time.time()

    feed_dict = fill_feed_dict(
      data_sets.train, images_placeholder, labels_placeholder)
    _, loss_value = sess.run([train_op, loss], feed_dict=feed_dict)

    duration = time.time() - start_time

    if step % 100 == 0:
      print('Step %d: loss = %.2f (%.3f sec)' % (step, loss_value, duration))
    
    if (step + 1) % 1000 == 0 or (step + 1) == max_steps:
      print('Test Data Eval:')
      do_eval(sess, eval_correct, images_placeholder, 
        labels_placeholder, data_sets.test)

  # saver.save(sess, checkpoint_file)

#====
#from tensorflow.python.tools import inspect_checkpoint as chkp
#chkp.print_tensors_in_checkpoint_file("model.ckpt", tensor_name='', all_tensors=True)


# But it is possible that backward nodes are not required int the graph before training... right?
# https://www.tensorflow.org/api_guides/python/meta_graph


sess = tf.InteractiveSession()
new_saver = tf.train.import_meta_graph(meta_file)
g = new_saver.export_meta_graph()
print(g)