Switch from model.predict(x) to model(x).numpy() for TF performance

FAQ.md

@@ -158,7 +158,7 @@ To check if the input values really are the same, you can print them, in Python
 input = ...
 print(input)
 print(input.shape)
-result = model.predict([input])
+result = model([input]).numpy()
 print(result)
 print(result.shape)  # result[0].shape in case of multiple output tensors
 ```

keras_export/convert_model.py

@@ -71,7 +71,7 @@ def get_model_input_layers(model):
 def measure_predict(model, data_in):
     """Returns output and duration in seconds"""
     start_time = datetime.datetime.now()
-    data_out = model.predict(data_in)
+    data_out = model(data_in).numpy()
     end_time = datetime.datetime.now()
     duration = end_time - start_time
     print('Forward pass took {} s.'.format(duration.total_seconds()))
@@ -558,11 +558,14 @@ def get_all_weights(model, prefix):
     layers = model.layers
     assert K.image_data_format() == 'channels_last'
     for layer in layers:
+        layer_type = type(layer).__name__
         for node in layer._inbound_nodes:
             if "training" in node.arguments.kwargs:
-                assert node.arguments.kwargs["training"] is not True, \
+                is_layer_with_accidental_training_flag = layer_type in ("CenterCrop", "Resizing")
+                has_training = node.arguments.kwargs["training"] is True
+                assert not has_training or is_layer_with_accidental_training_flag, \
                     "training=true is not supported, see https://github.com/Dobiasd/frugally-deep/issues/284"
-        layer_type = type(layer).__name__
+
         name = prefix + layer.name
         assert is_ascii(name)
         if name in result: