Upgrading your code to TensorFlow 2.0
Posted by Paige Bailey and Anna Revinskaya
TensorFlow 2.0 will include many API changes, such as reordering arguments, renaming symbols, and changing default values for parameters. Manually performing all of these modifications would be tedious, and prone to error. To streamline the changes, and to make your transition to TF 2.0 as seamless as possible, the TensorFlow engineering team has created a
tf_upgrade_v2 utility that will help transition legacy code to the new API.
tf_upgrade_v2 utility is included automatically with a pip install of TF 2.0, and will help accelerate your upgrade processes by converting existing TensorFlow 1.13 Python scripts to TensorFlow 2.0.
We have attempted to automate as many of the upgrade tasks as possible: however, there are still syntactical and stylistic changes that cannot be performed by the script.
Certain API symbols cannot be upgraded simply by using a string replacement. To ensure your code is still supported in TensorFlow 2.0, the upgrade script includes a
compat.v1 module. This module will replace calls of the form
tf.foo with equivalent
tf.compat.v1.foo references. It is recommended, however, to manually proofread such replacements and migrate them to new APIs in
tf.* namespace instead of
tf.compat.v1.* namespace as quickly as feasible.
Furthermore, due to module deprecations (for example,
tf.contrib), TensorFlow 2.0 will include changes that cannot be worked around by switching to
compat.v1. Upgrading code that uses these modules might require using an additional library (for e.g.
absl.flags) or switching to a package in
If you want to try upgrading your models from TensorFlow 1.12 to TensorFlow 2.0, follow the instructions below:
tf_upgrade_v2 is installed automatically by pip install for TensorFlow 1.13 and later (incl. the nightly 2.0 builds).
The upgrade script can be run on a single Python file:
tf_upgrade_v2 — infile foo.py — outfile foo-upgraded.py
You can also run it on a directory tree:
# upgrade the .py files and copy all the other files to the outtree
tf_upgrade_v2 — intree foo/ — outtree foo-upgraded/
# just upgrade the .py files
tf_upgrade_v2 — intree foo/ — outtree foo-upgraded/ — copyotherfiles False
The script also reports a list of detailed changes: for example, argument renames:
And any recommended manual checks:
All of this information is included in the
report.txt file that will be exported to your current directory. Once
tf_upgrade_v2 has run and exported your upgraded script, you can run your model and check to ensure your outputs are similar to TF 1.13:
- Do not update parts of your code manually before running this script. In particular, functions that have had reordered arguments like
tf.batch_to_spacecause the script to incorrectly add keyword arguments that mismap your existing code.
- This script does not reorder arguments. Instead, the script adds keyword arguments to functions that have their arguments reordered.
To report upgrade script bugs or make feature requests, file an issue on GitHub — and if you’re testing TensorFlow 2.0, we want to hear about it! Please feel free to join the TF 2.0 Testing community and send questions and discussion to email@example.com.