TFX cannot be installed in Colab since Colab runs on Python 3.10

[tanguycdls]

If the bug is related to a specific library below, please raise an issue in the
respective repo directly:

TensorFlow Data Validation Repo

TensorFlow Model Analysis Repo

TensorFlow Transform Repo

TensorFlow Serving Repo

System information

• Have I specified the code to reproduce the issue (Yes, No): yes
  !pip install tfx==1.13.0
• Environment in which the code is executed (e.g., Local(Linux/MacOS/Windows),
  Interactive Notebook, Google Cloud, etc):
  colab
• TensorFlow version:
• TFX Version: 1.13
• Python version:
• Python dependencies (from pip freeze output):

Describe the current behavior
TFX requires python up to 3.9, while colab runs on Python 3.10.
Describe the expected behavior
Be able to install TFX on colab.
Standalone code to reproduce the issue
!pip install tfx==1.13.0

Providing a bare minimum test case or step(s) to reproduce the problem will
greatly help us to debug the issue. If possible, please share a link to
Colab/Jupyter/any notebook.

Name of your Organization (Optional)
Tinyclues / Splio

Other info / logs

Include any logs or source code that would be helpful to diagnose the problem.
If including tracebacks, please include the full traceback. Large logs and files
should be attached.

[singhniraj08]

@tanguycdls,

We are aware of this issue and working on bumping up the python release to 3.10 for TFX. Meanwhile we tried a workaround of downgrading python version in colab environment but it fails once we restart the notebook. We will update this thread once we have an update. Thank you!

[singhniraj08]

@tanguycdls,

Meanwhile we are upgrading python support to 3.10, you can try using fallback runtime version to run the TFX tutorials. To enable that, once colab runtime is connected, You can use key combination (Ctrl + Shift + P) and type "Use fallback runtime version" and select it to use Python 3.9.16 version runtime. This is allow you to run TFX examples on fallback runtime.
Thanks.

[tanguycdls]

Helo @singhniraj08 thanks it works ! how long this fallback runtime is kept by Colab ?

[AnuarTB]

Wow, I didn't know about this feature. Thank you @singhniraj08! I will try to put out this tip on the tutorials website as well, after checking.

[AnuarTB]

CC: @rcrowe-google

This might be our temporary solution while we are migrating the libraries to Python 3.10.

[singhniraj08]

@AnuarTB, @tanguycdls,

As per this issue #colabtools/3572, the the fallback runtime version is available until mid-May.

[AnuarTB]

We are currently migrating everything, but the timeline is still not very clear. Will update if we find something new.

[GaneshACN]

any fix / progress about this Python 3.10 error for TFX DV library?

[rcrowe-google]

As a short term fix, this page has instructions for reverting to Python 3.8.3:

https://www.tensorflow.org/tfx/tutorials

[tanduong]

Do we have an update on this issue?

[AnuarTB]

We are actively working on migration. Unfortunately we cannot give definite ETA, but right now it is our most prioritized item to migrate TFX to Python 3.10.

[singhniraj08]

@tanguycdls,

Python 3.10 support is introduced in TFX 1.14.0 Release. Simply pip install tfx in colab should install TFX without any issues in Colab.
Please let us know if you have any issues. Thank you!

[github-actions]

This issue has been marked stale because it has no recent activity since 7 days. It will be closed if no further activity occurs. Thank you.

[AnuarTB]

As Python 3.10 is supported I am closing this issue.

[github-actions]

Are you satisfied with the resolution of your issue?
Yes
No

[raminmohammadi]

This issue still happens on colab with runtime Python 3.10.12, I am able to install TFX using pip install tfx. However, uppon importing tfx i get

`Using TensorFlow backend

ImportError Traceback (most recent call last)
in <cell line: 6>()
4 from absl import logging
5
----> 6 from tfx import v1 as tfx
7 from tfx.proto import example_gen_pb2, trainer_pb2
8 from tfx.orchestration.experimental.interactive.interactive_context import InteractiveContext

4 frames
/usr/local/lib/python3.10/dist-packages/tfx/components/tuner/executor.py in
20 from absl import logging
21 from keras_tuner.engine import base_tuner
---> 22 from keras_tuner.engine import trial
23 from tfx import types
24 from tfx.components.trainer import fn_args_utils

ImportError: cannot import name 'trial' from 'keras_tuner.engine' (/usr/local/lib/python3.10/dist-packages/keras_tuner/engine/init.py)

---

NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.

To view examples of installing some common dependencies, click the
"Open Examples" button below.
---------------------------------------------------------------------------`

Link to the colab: https://colab.research.google.com/drive/1ap8Gycu7s--mz0VAxp4W2DphAd1HW1yi?usp=sharing

[singhniraj08]

@raminmohammadi,

I am able to replicate this issue and the issue is because of keras-tuner package which fails to import from keras_tuner.engine import trial.
We are working on this issue. Meanwhile, you can degrade the keras-tuner version to make the tutorial work. Add the below lines of code after installing TFX in Colab. Thanks.

!pip uninstall keras-tuner
!pip install keras-tuner==1.3.5

[rcrowe-google]

The 1.4.2 release of Keras Tuner is now available on PyPI. Could you see if that fixes the problem for you?

[raminmohammadi]

@rcrowe-google it solved the problem.

[github-actions]

Are you satisfied with the resolution of your issue?
Yes
No