Error while using mobilenet weights

[pateldigant]

This code works fine for xception weighs but when i tried using mobilenet weights i get this error while trying to instantiate deeplab_model = Deeplabv3()
"ValueError: Layer #148 (named "image_pooling"), weight <tf.Variable 'image_pooling_1/kernel:0' shape=(1, 1, 2048, 256) dtype=float32_ref> has shape (1, 1, 2048, 256), but the saved weight has shape (256, 320, 1, 1).
"

[pateldigant]

This error was for the previous release.