Update DALI TensorFlow examples to work with 2.11 (#4554)

- updates RN50 example
- updates YOLO example
- updates EfficientDet example

Signed-off-by: Janusz Lisiecki <[email protected]>

Author: JanuszL

docs/examples/use_cases/tensorflow/efficientdet/model/utils/optimizers.py

@@ -27,10 +27,10 @@ def get_optimizer(params, *args):
 
     if params["optimizer"].lower() == "sgd":
         logging.info("Use SGD optimizer")
-        optimizer = tf.keras.optimizers.SGD(learning_rate, momentum=params["momentum"])
+        optimizer = tf.keras.optimizers.legacy.SGD(learning_rate, momentum=params["momentum"])
     elif params["optimizer"].lower() == "adam":
         logging.info("Use Adam optimizer")
-        optimizer = tf.keras.optimizers.Adam(learning_rate)
+        optimizer = tf.keras.optimizers.legacy.Adam(learning_rate)
     else:
         raise ValueError("optimizers should be adam or sgd")
     return optimizer

docs/examples/use_cases/tensorflow/resnet-n/nvutils/runner.py

@@ -19,16 +19,13 @@
 from distutils.version import StrictVersion
 
 import tensorflow as tf
-import tensorflow.keras as keras
+import keras
 import os
 import time
 import re
-import numpy as np
 import horovod.tensorflow.keras as hvd
 
-from tensorflow.python.keras.optimizer_v2 import (gradient_descent as
-                                                  gradient_descent_v2)
-from tensorflow.python.keras import backend
+from keras import backend
 print(tf.__version__)
 if StrictVersion(tf.__version__) > StrictVersion("2.1.0"):
   if StrictVersion(tf.__version__) >= StrictVersion("2.4.0"):

docs/examples/use_cases/tensorflow/resnet-n/nvutils/runner_ctl.py

@@ -20,19 +20,12 @@
 from distutils.version import StrictVersion
 
 import tensorflow as tf
-import tensorflow.keras as keras
-from tensorflow.python.ops import data_flow_ops
+import keras
 import os
-import sys
 import time
-import argparse
-import datetime
-import random
 import re
-import numpy as np
-from tensorflow.python.keras.optimizer_v2 import (gradient_descent as
-                                                  gradient_descent_v2)
-from tensorflow.python.keras import backend
+
+from keras import backend
 print(tf.__version__)
 if StrictVersion(tf.__version__) > StrictVersion("2.1.0"):
   if StrictVersion(tf.__version__) >= StrictVersion("2.4.0"):

docs/examples/use_cases/tensorflow/yolov4/src/train.py

@@ -118,7 +118,7 @@ def dataset_fn(input_context):
     with strategy.scope():
         model = YOLOv4Model()
         model.compile(
-            optimizer=tf.keras.optimizers.SGD(learning_rate=lr_fn)
+            optimizer=tf.keras.optimizers.legacy.SGD(learning_rate=lr_fn)
         )
 
     if start_weights:

qa/TL3_EfficientDet_convergence/test_tensorflow.sh

@@ -28,6 +28,4 @@ python train.py
     --ckpt_dir .                                                                                    \
     --output_filename out_weights_1.h5  2>&1 | tee $LOG
 
-popd
-
 CLEAN_AND_EXIT ${PIPESTATUS[0]}