Add estimator spec for model evaluation

model.py

@@ -22,8 +22,8 @@
 tf.logging.set_verbosity(tf.logging.INFO)
 
 # Training Parameters
-learning_rate_old = 1e-10
-num_steps = 9000
+start_lr = 1e-10
+training_steps = 12000
 batch_size = 128
 
 
@@ -95,35 +95,36 @@ def model_fn(features, labels, mode):
         logits=logits_train, labels=tf.cast(labels, dtype=tf.int32)))
 
     train_ops = None
+    acc_op = None
 
     # --------------------------------------------------------------------------
     # Optimize
     # --------------------------------------------------------------------------
     if mode == tf.estimator.ModeKeys.TRAIN:
         global_step = tf.train.get_global_step()
         learning_rate = tf.train.exponential_decay(
-            learning_rate_old, global_step=global_step,
+            start_lr, global_step=global_step,
             decay_steps=100, decay_rate=1.30)
         optimizer = tf.train.AdamOptimizer(learning_rate)
         train_ops = optimizer.minimize(loss, global_step=global_step)
         tf.summary.scalar("learning_rate", learning_rate)
         tf.summary.scalar("current_step", global_step)
         tf.summary.scalar("loss", loss)
 
-
-    # Evaluate the accuracy of the model
-    acc_op = tf.metrics.accuracy(labels=labels, predictions=pred_classes)
-
 
+    # Evaluate the accuracy of the model
+    if mode == tf.estimator.ModeKeys.EVAL:
+        acc_op = tf.metrics.accuracy(labels=labels, predictions=pred_classes)
+        return tf.estimator.EstimatorSpec(mode, loss=loss, eval_metric_ops={'accuracy': acc_op})
+
 
     # TF Estimators requires to return a EstimatorSpec, that specify
     # the different ops for training, evaluating, ...
     estim_specs = tf.estimator.EstimatorSpec(
         mode=mode,
         predictions=pred_classes,
         loss=loss,
-        train_op=train_ops,
-        eval_metric_ops={'accuracy': acc_op})
+        train_op=train_ops)
 
     return estim_specs
 
@@ -136,7 +137,7 @@ def model_fn(features, labels, mode):
     batch_size=batch_size, num_epochs=None, shuffle=True)
 
 # Train the Model
-model.train(input_fn, steps=num_steps)
+model.train(input_fn, steps=training_steps)
 
 # Evaluate the Model
 # Define the input function for evaluating
@@ -146,4 +147,6 @@ def model_fn(features, labels, mode):
 # Use the Estimator 'evaluate' method
 e = model.evaluate(input_fn)
 
+print("The test accuracy of the network: ", e['accuracy'])
+