Move resnet cifar to benchmark models.

PiperOrigin-RevId: 292007895

official/benchmark/keras_cifar_benchmark.py

@@ -24,7 +24,7 @@ import tensorflow as tf  # pylint: disable=g-bad-import-order
 
 from official.benchmark import keras_benchmark
 from official.utils.testing import benchmark_wrappers
-from official.vision.image_classification import resnet_cifar_main
+from official.benchmark.models import resnet_cifar_main
 
 MIN_TOP_1_ACCURACY = 0.929
 MAX_TOP_1_ACCURACY = 0.938

official/vision/image_classification/resnet_cifar_main.py → official/benchmark/models/resnet_cifar_main.py

@@ -18,17 +18,16 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 
-from absl import flags
 from absl import app as absl_app
+from absl import flags
 import tensorflow as tf
-
+from official.benchmark.models import resnet_cifar_model
 from official.utils.flags import core as flags_core
 from official.utils.logs import logger
 from official.utils.misc import distribution_utils
 from official.utils.misc import keras_utils
 from official.vision.image_classification import cifar_preprocessing
 from official.vision.image_classification import common
-from official.vision.image_classification import resnet_cifar_model
 
 
 LR_SCHEDULE = [  # (multiplier, epoch to start) tuples

official/vision/image_classification/resnet_cifar_test.py → official/benchmark/models/resnet_cifar_test.py

@@ -24,10 +24,10 @@ import tensorflow as tf
 
 from tensorflow.python.eager import context
 from tensorflow.python.platform import googletest
+from official.benchmark.models import resnet_cifar_main
 from official.utils.misc import keras_utils
 from official.utils.testing import integration
 from official.vision.image_classification import cifar_preprocessing
-from official.vision.image_classification import resnet_cifar_main
 
 
 class KerasCifarTest(googletest.TestCase):

official/vision/image_classification/README.md

@@ -11,13 +11,11 @@ For more information about other types of models, please refer to this
 ## ResNet
 
 Similar to the [estimator implementation](../../r1/resnet), the Keras
-implementation has code for both CIFAR-10 data and ImageNet data. The CIFAR-10
-version uses a ResNet56 model implemented in
-[`resnet_cifar_model.py`](./resnet_cifar_model.py), and the ImageNet version
-uses a ResNet50 model implemented in [`resnet_model.py`](./resnet_model.py).
+implementation has code for the ImageNet dataset. The ImageNet
+version uses a ResNet50 model implemented in
+[`resnet_model.py`](./resnet_model.py).
 
-To use
-either dataset, make sure that you have the latest version of TensorFlow
+Please make sure that you have the latest version of TensorFlow
 installed and
 [add the models folder to your Python path](/official/#running-the-models).
 
@@ -28,26 +26,6 @@ installed and
 * ResNet50 TFHub: [feature vector](https://tfhub.dev/tensorflow/resnet_50/feature_vector/1)
 and [classification](https://tfhub.dev/tensorflow/resnet_50/classification/1)
 
-### CIFAR-10
-
-Download and extract the CIFAR-10 data. You can use the following script:
-```bash
-python ../../r1/resnet/cifar10_download_and_extract.py
-```
-
-After you download the data, you can run the program by:
-
-```bash
-python resnet_cifar_main.py
-```
-
-If you did not use the default directory to download the data, specify the
-location with the `--data_dir` flag, like:
-
-```bash
-python resnet_cifar_main.py --data_dir=/path/to/cifar
-```
-
 ### ImageNet Training
 
 Download the ImageNet dataset and convert it to TFRecord format.