Move Keras Hook to use global step to resolve issues across epochs. (#7186)

* Move to global_step.

* Hook to use global_step.

* fix comment start step 1 not step 0.

* remove hack used for testing.

* Add docstring.

official/resnet/keras/keras_cifar_benchmark.py

@@ -22,10 +22,8 @@ import time
 from absl import flags
 import tensorflow as tf  # pylint: disable=g-bad-import-order
 
-from official.resnet import cifar10_main as cifar_main
 from official.resnet.keras import keras_benchmark
 from official.resnet.keras import keras_cifar_main
-from official.resnet.keras import keras_common
 
 MIN_TOP_1_ACCURACY = 0.925
 MAX_TOP_1_ACCURACY = 0.938

official/utils/misc/keras_utils.py

@@ -46,41 +46,37 @@ class TimeHistory(tf.keras.callbacks.Callback):
     Args:
       batch_size: Total batch size.
       log_steps: Interval of time history logs.
-
     """
     self.batch_size = batch_size
     super(TimeHistory, self).__init__()
     self.log_steps = log_steps
+    self.global_steps = 0
 
-    # Logs start of step 0 then end of each step based on log_steps interval.
+    # Logs start of step 1 then end of each step based on log_steps interval.
     self.timestamp_log = []
 
-  def on_train_begin(self, logs=None):
-    self.record_batch = True
-
   def on_train_end(self, logs=None):
     self.train_finish_time = time.time()
 
   def on_batch_begin(self, batch, logs=None):
-    if self.record_batch:
-      timestamp = time.time()
-      self.start_time = timestamp
-      self.record_batch = False
-      if batch == 0:
-        self.timestamp_log.append(BatchTimestamp(batch, timestamp))
+    self.global_steps += 1
+    if self.global_steps == 1:
+      self.start_time = time.time()
+      self.timestamp_log.append(BatchTimestamp(self.global_steps,
+                                               self.start_time))
 
   def on_batch_end(self, batch, logs=None):
-    if batch % self.log_steps == 0:
+    """Records elapse time of the batch and calculates examples per second."""
+    if self.global_steps % self.log_steps == 0:
       timestamp = time.time()
       elapsed_time = timestamp - self.start_time
       examples_per_second = (self.batch_size * self.log_steps) / elapsed_time
-      if batch != 0:
-        self.record_batch = True
-        self.timestamp_log.append(BatchTimestamp(batch, timestamp))
-        tf.compat.v1.logging.info(
-            "BenchmarkMetric: {'num_batches':%d, 'time_taken': %f,"
-            "'examples_per_second': %f}" %
-            (batch, elapsed_time, examples_per_second))
+      self.timestamp_log.append(BatchTimestamp(self.global_steps, timestamp))
+      tf.compat.v1.logging.info(
+          "BenchmarkMetric: {'global step':%d, 'time_taken': %f,"
+          "'examples_per_second': %f}" %
+          (self.global_steps, elapsed_time, examples_per_second))
+      self.start_time = timestamp
 
 
 def get_profiler_callback(model_dir, profile_steps, enable_tensorboard):