(Fixed) ModuleNotFoundError: No Module Named ‘Keras’

5/5 - (2 votes)

Quick Fix: Python raises the ImportError: No module named 'keras' when it cannot find the TensorFlow library that also contains the keras module. To fix it, install TensorFlow using PIP and import Keras using from tensorflow import keras, and not import keras.

Here’s how to install TensorFlow if you haven’t already:

pip install tensorflow
pip3 install tensorflow

Here’s how you correctly import Keras in your Python script

from tensorflow import keras

Alternatively, you may have different Python versions on your computer and keras is not installed for the version you’re using. I’ll show you how to fix this issue below after the “Problem Formulation”. 👇

Problem Formulation

You’ve just learned about the awesome capabilities of the keras library and you want to try it out, so you start your code with the following statement:

import keras

This is supposed to import the Keras library into your (virtual) environment. However, it only raises the ImportError: No module named keras or ModuleNotFoundError: No module named 'keras':

>>> import keras
Traceback (most recent call last):
  File "<pyshell#6>", line 1, in <module>
    import keras
ModuleNotFoundError: No module named 'keras'

Note that you should actually use the following statement to import Keras from the TensorFlow library:

>>> from tensorflow import keras
Traceback (most recent call last):
  File "<pyshell#2>", line 1, in <module>
    from tensorflow import keras
ModuleNotFoundError: No module named 'tensorflow'

Solution Idea 1: Install Library TensorFlow

The most likely reason is that Python doesn’t provide keras in its standard library. You shouldn’t install it with pip install keras either. Instead, you need to install the TensorFlow library using Python’s package manager pip. Make sure pip is installed on your machine.

To fix this error, you can run the following command in your shell:

pip install tensorflow

This simple command installs tensorflow in your virtual environment on Windows, Linux, and MacOS. It assumes that your pip version is updated. If it isn’t, use the following two commands in your terminal, command line, or shell (there’s no harm in doing it anyways):

python -m pip install --upgrade pip
pip install tensorflow

In both code examples above, you may also want to replace pip with pip3 if you have multiple Python versions and you want to install TensorFlow for Python 3:

pip3 install tensorflow

Also, you may want to check out the official installation guide from TensorFlow, depending on the concrete environment you’re using and the goals you’re pursuing with your TensorFlow installation.

For instance, some people want to install TensorFlow in their Anacondas environment and also enable GPU support. If this is you, you should first install Miniconda (a free installer for conda), create a conda virtual environment, and install TensorFlow in that environment using PIP.

Solution Idea 2: Fix the Path

The error might persist even after you have installed the tensorflow library. This likely happens because pip is installed but doesn’t reside in the path you can use. Although pip may be installed on your system the script is unable to locate it. Therefore, it is unable to install the library using pip in the correct path.

To fix the problem with the path in Windows follow the steps given next.

Step 1: Open the folder where you installed Python by opening the command prompt and typing where python

Step 2: Once you have opened the Python folder, browse and open the Scripts folder and copy its location. Also verify that the folder contains the pip file.

Step 3: Now open the Scripts directory in the command prompt using the cd command and the location that you copied previously.

Step 4: Now install the library using pip install tensorflow command. Here’s an analogous example:

After having followed the above steps, execute our script once again. And you should get the desired output.

Understanding the “import” Statement

from tensorflow import keras

In Python, the import statement serves two main purposes:

Search the module by its name, load it, and initialize it.
Define a name in the local namespace within the scope of the import statement. This local name is then used to reference the accessed module throughout the code.

If you experience the following variant of the error, here’s an easy fix: 👇

ModuleNotFoundError: No module named ‘tensorflow.keras’

You seem to be trying to import the tensorflow.keras module in Python, but Python can’t find it. This could be due to a few reasons:

(1) TensorFlow isn’t installed: If TensorFlow isn’t installed in your current Python environment, you can install it using pip:

pip install tensorflow

Or, if you are using a Jupyter notebook and the above command doesn’t work, try:

!pip install tensorflow

(2) You are using an incorrect import statement: You might be trying to import Keras from TensorFlow in a wrong way. The correct way to import Keras from TensorFlow is:

from tensorflow import keras

(3) Your TensorFlow installation is broken or incomplete: If TensorFlow is installed but the installation is broken, you might need to uninstall it first before reinstalling. Here’s how you can do it:

pip uninstall tensorflow
pip install tensorflow

(4) Your Python environment is not activated or not correctly set: If you’re using a virtual environment or a conda environment, you need to make sure that the environment is activated and that TensorFlow is installed in that environment.

(5) TensorFlow version compatibility issue: As of my knowledge cut-off in September 2021, TensorFlow 2.x includes Keras in the tensorflow.keras module. If you’re using TensorFlow 1.x, you might need to upgrade to a later version. You can upgrade TensorFlow using pip:

pip install --upgrade tensorflow

Remember to replace pip with pip3 and add python -m at the start if you’re using Python 3 and having issues with the pip command.

What’s the Difference Between ImportError and ModuleNotFoundError?

What’s the difference between ImportError and ModuleNotFoundError?

Python defines an error hierarchy, so some error classes inherit from other error classes. In our case, the ModuleNotFoundError is a subclass of the ImportError class.

You can see this in this screenshot from the docs:

You can also check this relationship using the issubclass() built-in function:

>>> issubclass(ModuleNotFoundError, ImportError)
True

Specifically, Python raises the ModuleNotFoundError if the module (e.g., keras or tensorflow) cannot be found. If it can be found, there may be a problem loading the module or some specific files within the module. In those cases, Python would raise an ImportError.

If an import statement cannot import a module, it raises an ImportError. This may occur because of a faulty installation or an invalid path. In Python 3.6 or newer, this will usually raise a ModuleNotFoundError.

How to Fix “ModuleNotFoundError: No module named ‘keras'” in PyCharm

If you create a new Python project in PyCharm and try to import the keras library, it’ll raise the following error message:

Traceback (most recent call last):
  File "C:/Users/.../main.py", line 1, in <module>
    import keras
ModuleNotFoundError: No module named 'keras'

Process finished with exit code 1

The reason is that each PyCharm project, per default, creates a virtual environment in which you can install custom Python modules. But the virtual environment is initially empty—even if you’ve already installed tensorflow on your computer!

Here’s a screenshot exemplifying this for the pandas library. It’ll look similar for keras.

The fix is simple: Use the PyCharm installation tooltips to install Pandas in your virtual environment—two clicks and you’re good to go!

First, right-click on the pandas text in your editor:

Second, click “Show Context Actions” in your context menu. In the new menu that arises, click "Install Pandas" and wait for PyCharm to finish the installation.

The code will run after your installation completes successfully.

As an alternative, you can also open the Terminal tool at the bottom and type:

$ pip install tensorflow

If this doesn’t work, you may want to set the Python interpreter to another version using the following tutorial: https://www.jetbrains.com/help/pycharm/2016.1/configuring-python-interpreter-for-a-project.html

You can also manually install a new library such as tensorflow in PyCharm using the following procedure:

Open File > Settings > Project from the PyCharm menu.
Select your current project.
Click the Python Interpreter tab within your project tab.
Click the small + symbol to add a new library to the project.
Now type in the library to be installed, in your example Pandas, and click Install Package.
Wait for the installation to terminate and close all popup windows.

Here’s an analogous example:

Here’s a full guide on how to install a library on PyCharm.

How to Install a Library on PyCharm

Chris

While working as a researcher in distributed systems, Dr. Christian Mayer found his love for teaching computer science students.

To help students reach higher levels of Python success, he founded the programming education website Finxter.com that has taught exponential skills to millions of coders worldwide. He’s the author of the best-selling programming books Python One-Liners (NoStarch 2020), The Art of Clean Code (NoStarch 2022), and The Book of Dash (NoStarch 2022). Chris also coauthored the Coffee Break Python series of self-published books. He’s a computer science enthusiast, freelancer, and owner of one of the top 10 largest Python blogs worldwide.

His passions are writing, reading, and coding. But his greatest passion is to serve aspiring coders through Finxter and help them to boost their skills. You can join his free email academy here.