bug fix

official/resnet/keras/keras_cifar_main.py

@@ -112,6 +112,7 @@ def run_cifar_with_keras(flags_obj):
   Raises:
     ValueError: If fp16 is passed as it is not currently supported.
   """
+  print(">>>>>>>>>>>>>>>>>>>>> eager: ", flags_obj.enable_eager)
   if flags_obj.enable_eager:
     tf.enable_eager_execution()
 
@@ -199,13 +200,13 @@ def run_cifar_with_keras(flags_obj):
                                verbose=1)
 
   print('Test loss:', eval_output[0])
-  stats = keras_common.analyze_eval_result(eval_output)
+  stats = keras_common.analyze_fit_and_eval_result(history, eval_output)
 
   return stats
 
 
 def define_keras_cifar_flags():
-  flags.DEFINE_boolean(name='enable_eager', default=False, help='Enable eager?')
+  flags.DEFINE_boolean(name='enable_eager', default=True, help='Enable eager?')
 
 
 def main(_):

official/resnet/keras/keras_common.py

@@ -138,6 +138,7 @@ def get_dist_strategy():
     print('Not using distribution strategies.')
     strategy = None
   else:
+    print(">>>>>>>>>>>>>>>>>>strategy!!!!!!! ", FLAGS.num_gpus)
     strategy = distribution_utils.get_distribution_strategy(
         num_gpus=FLAGS.num_gpus)
 
@@ -158,7 +159,7 @@ def get_fit_callbacks(learning_rate_schedule_fn):
 
   return time_callback, tensorboard_callback, lr_callback
 
-def analyze_eval_result(eval_output):
+def analyze_fit_and_eval_result(history, eval_output):
   stats = {}
   stats['accuracy_top_1'] = eval_output[1]
   stats['eval_loss'] = eval_output[0]

official/resnet/keras/keras_imagenet_main.py

@@ -178,23 +178,24 @@ def run_imagenet_with_keras(flags_obj):
   num_eval_steps = (imagenet_main._NUM_IMAGES['validation'] //
                   flags_obj.batch_size)
 
-  model.fit(train_input_dataset,
-            epochs=flags_obj.train_epochs,
-            steps_per_epoch=steps_per_epoch,
-            callbacks=[
-              time_callback,
-              lr_callback,
-              tensorboard_callback
-            ],
-            validation_steps=num_eval_steps,
-            validation_data=eval_input_dataset,
-            verbose=1)
+  history = model.fit(train_input_dataset,
+                      epochs=flags_obj.train_epochs,
+                      steps_per_epoch=steps_per_epoch,
+                      callbacks=[
+                        time_callback,
+                        lr_callback,
+                        tensorboard_callback
+                      ],
+                      validation_steps=num_eval_steps,
+                      validation_data=eval_input_dataset,
+                      verbose=1)
 
   eval_output = model.evaluate(eval_input_dataset,
                                steps=num_eval_steps,
                                verbose=1)
+
   print('Test loss:', eval_output[0])
-  stats = keras_common.analyze_eval_result(eval_output)
+  stats = keras_common.analyze_fit_and_eval_result(history, eval_output)
 
   return stats