Deep Learning with TensorFlow

Limitations

The tf_upgrade.py script has some limitations. For example, you must manually fix any instances of tf.reverse(). This is not that hard since the tf_upgrade.py script will warn you about the tf.reverse() in the report.txt file. For the reordered arguments, tf_upgrade.py tries to minimally reformat your code. However, the actual argument order cannot be changed automatically. For another example, constructions such as tf.get_variable_scope().reuse_variables() will likely not work. Therefore, we recommend deleting those lines and replacing them with the following:  tf.variable_scope(tf.get_variable_scope(), reuse=True).