def resnet_model_fn_helper(self, mode):
"""Tests that the EstimatorSpec is given the appropriate arguments."""
tf.train.create_global_step()
features, labels = self.input_fn()
spec = imagenet_main.resnet_model_fn(
features, labels, mode, {
'resnet_size': 50,
'data_format': 'channels_last',
'batch_size': _BATCH_SIZE,
})
predictions = spec.predictions
self.assertAllEqual(predictions['probabilities'].shape,
(_BATCH_SIZE, _LABEL_CLASSES))
self.assertEqual(predictions['probabilities'].dtype, tf.float32)
self.assertAllEqual(predictions['classes'].shape, (_BATCH_SIZE, ))
self.assertEqual(predictions['classes'].dtype, tf.int64)
if mode != tf.estimator.ModeKeys.PREDICT:
loss = spec.loss
self.assertAllEqual(loss.shape, ())
self.assertEqual(loss.dtype, tf.float32)
if mode == tf.estimator.ModeKeys.EVAL:
eval_metric_ops = spec.eval_metric_ops
self.assertAllEqual(eval_metric_ops['accuracy'][0].shape, ())
self.assertAllEqual(eval_metric_ops['accuracy'][1].shape, ())
self.assertEqual(eval_metric_ops['accuracy'][0].dtype, tf.float32)
self.assertEqual(eval_metric_ops['accuracy'][1].dtype, tf.float32)
def resnet_model_fn_helper(self, mode):
"""Tests that the EstimatorSpec is given the appropriate arguments."""
tf.train.create_global_step()
features, labels = self.input_fn()
spec = imagenet_main.resnet_model_fn(features, labels, mode)
predictions = spec.predictions
self.assertAllEqual(predictions['probabilities'].shape,
(FLAGS.train_batch_size, _LABEL_CLASSES))
self.assertEqual(predictions['probabilities'].dtype, tf.float32)
self.assertAllEqual(predictions['classes'].shape, (FLAGS.train_batch_size,))
self.assertEqual(predictions['classes'].dtype, tf.int64)
if mode != tf.estimator.ModeKeys.PREDICT:
loss = spec.loss
self.assertAllEqual(loss.shape, ())
self.assertEqual(loss.dtype, tf.float32)
if mode == tf.estimator.ModeKeys.EVAL:
eval_metric_ops = spec.eval_metric_ops
self.assertAllEqual(eval_metric_ops['accuracy'][0].shape, ())
self.assertAllEqual(eval_metric_ops['accuracy'][1].shape, ())
self.assertEqual(eval_metric_ops['accuracy'][0].dtype, tf.float32)
self.assertEqual(eval_metric_ops['accuracy'][1].dtype, tf.float32)