TensorFlow Model Analysis (TFMA) is a library for evaluating TensorFlow models. It allows users to evaluate their models on large amounts of data in a distributed manner, using the same metrics defined in their trainer. These metrics can be computed over different slices of data and visualized in Jupyter notebooks.
Caution: TFMA may introduce backwards incompatible changes before version 1.0.
The recommended way to install TFMA is using the PyPI package:
pip install tensorflow-model-analysis
pip install from https://pypi-nightly.tensorflow.org
pip install -i https://pypi-nightly.tensorflow.org/simple tensorflow-model-analysis
pip install from the HEAD of the git:
pip install git+https://github.com/tensorflow/model-analysis.git#egg=tensorflow_model_analysis
pip install from a released version directly from git:
pip install git+https://github.com/tensorflow/[email protected]#egg=tensorflow_model_analysis
If you have cloned the repository locally, and want to test your local change, pip install from a local folder.
pip install -e $FOLDER_OF_THE_LOCAL_LOCATION
Note that protobuf must be installed correctly for the above option since it is building TFMA from source and it requires protoc and all of its includes reference-able. Please see protobuf install instruction for see the latest install instructions.
Currently, TFMA requires that TensorFlow is installed but does not have an explicit dependency on the TensorFlow PyPI package. See the TensorFlow install guides for instructions.
To build from source follow the following steps:
Install the protoc as per the link mentioned: protoc
Create a virtual environment by running the commands
python3 -m venv <virtualenv_name>
source <virtualenv_name>/bin/activate
pip3 install setuptools wheel
git clone https://github.com/tensorflow/model-analysis.git
cd model-analysis
python3 setup.py bdist_wheel
This will build the TFMA wheel in the dist directory. To install the wheel from dist directory run the commands
cd dist
pip3 install tensorflow_model_analysis-<version>-py3-none-any.whl
As of writing, because of pypa/pip#9187, pip install
might never finish. In that case, you should revert pip to version 19 instead of
20: pip install "pip<20".
Using a JupyterLab extension requires installing dependencies on the command line. You can do this within the console in the JupyterLab UI or on the command line. This includes separately installing any pip package dependencies and JupyterLab labextension plugin dependencies, and the version numbers must be compatible. JupyterLab labextension packages refer to npm packages (eg, tensorflow_model_analysis.
The examples below use 0.32.0. Check available versions below to use the latest.
pip install tensorflow_model_analysis==0.32.0
jupyter labextension install [email protected]
pip install jupyterlab_widgets==1.0.0pip install tensorflow_model_analysis==0.32.0
jupyter labextension install [email protected]
jupyter labextension install @jupyter-widgets/jupyterlab-manager@2pip install tensorflow_model_analysis==0.32.0
jupyter labextension install [email protected]
jupyter labextension install @jupyter-widgets/[email protected]To enable TFMA visualization in the classic Jupyter Notebook (either through
jupyter notebook or
through the JupyterLab UI),
you'll also need to run:
jupyter nbextension enable --py widgetsnbextension
jupyter nbextension enable --py tensorflow_model_analysisNote: If Jupyter notebook is already installed in your home directory, add
--user to these commands. If Jupyter is installed as root, or using a virtual
environment, the parameter --sys-prefix might be required.
If you want to build TFMA from source and use the UI in JupyterLab, you'll need
to make sure that the source contains valid version numbers. Check that the
Python package version number and npm package version number are exactly the
same, and that both valid version numbers (eg, remove the -dev suffix).
Check pip packages:
pip listCheck JupyterLab extensions:
jupyter labextension list # for JupyterLab
jupyter nbextension list # for classic Jupyter NotebookTFMA notebook extension can be built into a standalone HTML file that also bundles data into the HTML file. See the Jupyter Widgets docs on embed_minimal_html.
Kubeflow Pipelines includes integrations that embed the TFMA notebook extension (code). This integration relies on network access at runtime to load a variant of the JavaScript build published on unpkg.com (see config and loader code).
TensorFlow is required.
Apache Beam is required; it's the way that efficient distributed computation is supported. By default, Apache Beam runs in local mode but can also run in distributed mode using Google Cloud Dataflow and other Apache Beam runners.
Apache Arrow is also required. TFMA uses Arrow to represent data internally in order to make use of vectorized numpy functions.
For instructions on using TFMA, see the get started guide.
The following table is the TFMA package versions that are compatible with each other. This is determined by our testing framework, but other untested combinations may also work.
| tensorflow-model-analysis | apache-beam[gcp] | pyarrow | tensorflow | tensorflow-metadata | tfx-bsl |
|---|---|---|---|---|---|
| GitHub master | 2.36.0 | 5.0.0 | nightly (1.x/2.x) | 1.7.0 | 1.7.0 |
| 0.38.0 | 2.36.0 | 5.0.0 | 1.15.5 / 2.8 | 1.7.0 | 1.7.0 |
| 0.37.0 | 2.35.0 | 5.0.0 | 1.15.5 / 2.7 | 1.6.0 | 1.6.0 |
| 0.36.0 | 2.34.0 | 5.0.0 | 1.15.5 / 2.7 | 1.5.0 | 1.5.0 |
| 0.35.0 | 2.33.0 | 5.0.0 | 1.15 / 2.6 | 1.4.0 | 1.4.0 |
| 0.34.1 | 2.32.0 | 2.0.0 | 1.15 / 2.6 | 1.2.0 | 1.3.0 |
| 0.34.0 | 2.31.0 | 2.0.0 | 1.15 / 2.6 | 1.2.0 | 1.3.1 |
| 0.33.0 | 2.31.0 | 2.0.0 | 1.15 / 2.5 | 1.2.0 | 1.2.0 |
| 0.32.1 | 2.29.0 | 2.0.0 | 1.15 / 2.5 | 1.1.0 | 1.1.1 |
| 0.32.0 | 2.29.0 | 2.0.0 | 1.15 / 2.5 | 1.1.0 | 1.1.0 |
| 0.31.0 | 2.29.0 | 2.0.0 | 1.15 / 2.5 | 1.0.0 | 1.0.0 |
| 0.30.0 | 2.28.0 | 2.0.0 | 1.15 / 2.4 | 0.30.0 | 0.30.0 |
| 0.29.0 | 2.28.0 | 2.0.0 | 1.15 / 2.4 | 0.29.0 | 0.29.0 |
| 0.28.0 | 2.28.0 | 2.0.0 | 1.15 / 2.4 | 0.28.0 | 0.28.0 |
| 0.27.0 | 2.27.0 | 2.0.0 | 1.15 / 2.4 | 0.27.0 | 0.27.0 |
| 0.26.1 | 2.28.0 | 0.17.0 | 1.15 / 2.3 | 0.26.0 | 0.26.0 |
| 0.26.0 | 2.25.0 | 0.17.0 | 1.15 / 2.3 | 0.26.0 | 0.26.0 |
| 0.25.0 | 2.25.0 | 0.17.0 | 1.15 / 2.3 | 0.25.0 | 0.25.0 |
| 0.24.3 | 2.24.0 | 0.17.0 | 1.15 / 2.3 | 0.24.0 | 0.24.1 |
| 0.24.2 | 2.23.0 | 0.17.0 | 1.15 / 2.3 | 0.24.0 | 0.24.0 |
| 0.24.1 | 2.23.0 | 0.17.0 | 1.15 / 2.3 | 0.24.0 | 0.24.0 |
| 0.24.0 | 2.23.0 | 0.17.0 | 1.15 / 2.3 | 0.24.0 | 0.24.0 |
| 0.23.0 | 2.23.0 | 0.17.0 | 1.15 / 2.3 | 0.23.0 | 0.23.0 |
| 0.22.2 | 2.20.0 | 0.16.0 | 1.15 / 2.2 | 0.22.2 | 0.22.0 |
| 0.22.1 | 2.20.0 | 0.16.0 | 1.15 / 2.2 | 0.22.2 | 0.22.0 |
| 0.22.0 | 2.20.0 | 0.16.0 | 1.15 / 2.2 | 0.22.0 | 0.22.0 |
| 0.21.6 | 2.19.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.3 |
| 0.21.5 | 2.19.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.3 |
| 0.21.4 | 2.19.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.3 |
| 0.21.3 | 2.17.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.0 |
| 0.21.2 | 2.17.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.0 |
| 0.21.1 | 2.17.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.0 |
| 0.21.0 | 2.17.0 | 0.15.0 | 1.15 / 2.1 | 0.21.0 | 0.21.0 |
| 0.15.4 | 2.16.0 | 0.15.0 | 1.15 / 2.0 | n/a | 0.15.1 |
| 0.15.3 | 2.16.0 | 0.15.0 | 1.15 / 2.0 | n/a | 0.15.1 |
| 0.15.2 | 2.16.0 | 0.15.0 | 1.15 / 2.0 | n/a | 0.15.1 |
| 0.15.1 | 2.16.0 | 0.15.0 | 1.15 / 2.0 | n/a | 0.15.0 |
| 0.15.0 | 2.16.0 | 0.15.0 | 1.15 | n/a | n/a |
| 0.14.0 | 2.14.0 | n/a | 1.14 | n/a | n/a |
| 0.13.1 | 2.11.0 | n/a | 1.13 | n/a | n/a |
| 0.13.0 | 2.11.0 | n/a | 1.13 | n/a | n/a |
| 0.12.1 | 2.10.0 | n/a | 1.12 | n/a | n/a |
| 0.12.0 | 2.10.0 | n/a | 1.12 | n/a | n/a |
| 0.11.0 | 2.8.0 | n/a | 1.11 | n/a | n/a |
| 0.9.2 | 2.6.0 | n/a | 1.9 | n/a | n/a |
| 0.9.1 | 2.6.0 | n/a | 1.10 | n/a | n/a |
| 0.9.0 | 2.5.0 | n/a | 1.9 | n/a | n/a |
| 0.6.0 | 2.4.0 | n/a | 1.6 | n/a | n/a |
Please direct any questions about working with TFMA to Stack Overflow using the tensorflow-model-analysis tag.