How to downgrade tensorflow version in colab?

From the moment we opened our eyes and stepped into this world, we were always surrounded by technology. Technology has now become our past, present and future. We have become dependent and more reliant on the technology because of its continuous advancements. One such emerging technology which is currently dominating all of us is the “Machine Learning” and its vast number of libraries and frameworks used in it.

One such intimate connection is of Machine Learning and Tensorflow. TensorFlow, developed by Google, is one of the most popular and powerful libraries as it offers a rich ecosystem of tools, libraries, and resources for creating and implementing machine learning models. However, there maybe some times when you need to switch or downgrade the version of the TensorFlow as per the requirements of your project. Here, we will use Google Colab, which is a cloud-based Jupyter Notebook environment, for working with the tensorflow as it already comes pre-installed with current version of TensorFlow.

In this article, we will walk through the steps on how to downgrade TensorFlow version in Colab. But, before directly jumping to downgrade or switch between the TensorFlow versions, let’s first understand why you may want to do so.

WHY DOWNGRADE TENSORFLOW VERSION IN COLAB?

1. Compatibility Issues: Your code or project may be written in an older version, but not compatible with the newer version. By downgrading your TensorFlow, you can make sure that your code will run without any errors or problems.

2. Dependencies: You may be using some third-party library, framework, or package that is not yet compatible with the new TensorFlow. By downgrading, you can maintain compatibility with those dependencies.

3. Legacy Projects: If you are working on an old machine learning/deep learning project, you may want to downgrade your TensorFlow to that version to make sure that the results are consistent and reproducible.

4. Experimentations: You may want to experiment with different versions to see how they perform or behave for your task. In that case, you can easily test different versions.

5. Resource Constraints: In some instances, the newer TensorFlow version may consume more memory or compute resources than what is available in the Colab environment. Instead, if you are working with a lighter version, you will be able to work within the resource constraints of the newer version.

6. Stability: The newer TensorFlow releases may still contain some bugs or some experimental features that have not been tested thoroughly. If you are looking for a stable environment, you may want to use an older, established version.

STEPS TO DOWNGRADE TENSORFLOW VERSION IN COLAB:

Step 1: CHECK THE CURRENT VERSION OF TENSORFLOW INSTALLED IN COLAB

Before you begin, check for the current version of TensorFlow installed in Colab. To do so, use the command given below

import tensorflow as tf
print(tf.__version__)

This will show the current version of TensorFlow installed in your Colab.

Output:

2.13.0

As shown above, mu current version is 2.13.0

Step 2: UNINSTALL THE CURRENT VERSION OF TENSORFLOW

In order to downgrade to an older version or switch from the current version of TensorFlow to another desired version, you need to first uninstall the current TensorFlow version installed in Colab. To uninstall the TensorFlow, run the following command in the code cell given below

!pip uninstall tensorflow

Once you run the above command in the code cell, it will uninstall the current version of TensorFlow from your Colab environment.

To do
Step 3: INSTALL THE CURRENT TENSORFLOW VERSION

In the next step, you need to install the new desired version of TensorFlow in Colab. For performing this, use the following command

!pip install tensorflow==<version>

Replace <version> with the desired version of TensorFlow that you want to be installed of your choice. For example, if you want to install the TensorFlow version 2.1.0, then write the same in the space provided.

!pip install tensorflow==2.1.0

This will install the version 2.1.0

Step 4: VERIFY THE TENSORFLOW VERSION

Once you are done with the installation, the next is its verification. You need to check whether the new installed version of TensorFlow successfully matches with your chosen version or not. To do this, run the following command

import tensorflow as tf
print(tf.__version__)

Output:

2.10.0

This will print the TensorFlow version that is currently installed in your Colab. If it matches with the version you installed in the previous step, then you have successfully installed TensorFlow.

You may need to restart the Colab runtime to use the downgraded TensorFlow version. You can do this by clicking on “Runtime” in the Colab menu and selecting “Restart runtime.” After restarting the runtime, you should be able to use the downgraded TensorFlow version in your Colab notebook.

Keep in mind that downgrading TensorFlow may affect compatibility with certain libraries or code that relies on newer TensorFlow features. Be sure to test your code to ensure it works as expected with the downgraded version.