sat

example_mnist.py

@@ -60,7 +60,7 @@ def get_model(input, label):
 
     y = one_hot(label, NUM_CLASS)
     cost = tf.nn.softmax_cross_entropy_with_logits(fc1, y)
-    cost = tf.reduce_sum(cost, name='cost')
+    cost = tf.reduce_mean(cost, name='cost')
 
     tf.scalar_summary(cost.op.name, cost)
     return prob, cost
@@ -97,14 +97,16 @@ def main():
         keep_prob = G.get_tensor_by_name('dropout_prob:0')
         with sess.as_default():
             for epoch in count(1):
+                running_cost = StatCounter()
                 for (img, label) in dataset_train.get_data():
                     feed = {input_var: img,
                             label_var: label,
                             keep_prob: 0.5}
 
                     _, cost_value = sess.run([train_op, cost], feed_dict=feed)
+                    running_cost.feed(cost_value)
 
-                print('Epoch %d: last batch cost = %.2f' % (epoch, cost_value))
+                print('Epoch %d: avg cost = %.2f' % (epoch, running_cost.average))
                 summary_str = summary_op.eval(feed_dict=feed)
                 summary_writer.add_summary(summary_str, epoch)
 

utils/extension.py

@@ -65,7 +65,7 @@ class OnehotClassificationValidation(PeriodicExtension):
 
     def _trigger(self):
         cnt = 0
-        cnt_correct = 0
+        correct_stat = Accuracy()
         sess = tf.get_default_session()
         cost_sum = 0
         for (img, label) in self.ds.get_data():
@@ -75,12 +75,12 @@ class OnehotClassificationValidation(PeriodicExtension):
             cnt += img.shape[0]
             correct, cost = sess.run([self.nr_correct_var, self.cost_var],
                                     feed_dict=feed)
-            cnt_correct += correct
-            cost_sum += cost
+            correct_stat.feed(correct, cnt)
+            cost_sum += cost * cnt
         cost_sum /= cnt
         # TODO write to summary?
         print "After epoch {}: acc={}, cost={}".format(
-            self.epoch_num, cnt_correct * 1.0 / cnt, cost_sum)
+            self.epoch_num, correct_stat.accuracy, cost_sum)
 
 
 class PeriodicSaver(PeriodicExtension):

utils/stat.py

+#!/usr/bin/env python2
+# -*- coding: UTF-8 -*-
+# File: stat.py
+# Author: Yuxin Wu <ppwwyyxx@gmail.com>
+import numpy as np
+
+
+class StatCounter(object):
+    def __init__(self):
+        self.values = []
+
+    def feed(self, v):
+        self.values.append(v)
+
+    @property
+    def average(self):
+        return np.mean(self.values)
+
+    @property
+    def sum(self):
+        return np.sum(self.values)
+
+class Accuracy(object):
+    def __init__(self):
+        self.tot = 0
+        self.corr = 0
+
+    def feed(self, corr, tot=1):
+        self.tot += tot
+        self.corr += corr
+
+    @property
+    def accuracy(self):
+        if self.tot < 0.001:
+            return 0
+        return self.corr * 1.0 / self.tot
+