TensorFlow Model Analysis

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.

Installation

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/model-analysis.git@v0.21.3#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.

Build TFMA from source

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

To enable TFMA visualization in Jupyter Notebook:

  jupyter nbextension enable --py widgetsnbextension
  jupyter nbextension enable --py tensorflow_model_analysis

Note: 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.

Jupyter Lab

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.

The examples below use 0.27.0. Check available versions below to use the latest.

Jupyter Lab 1.2.x

pip install tensorflow_model_analysis==0.27.0
jupyter labextension install tensorflow_model_analysis@0.27.0
jupyter labextension install @jupyter-widgets/jupyterlab-manager@1.1

Jupyter Lab 2

pip install tensorflow_model_analysis==0.27.0
jupyter labextension install tensorflow_model_analysis@0.27.0
jupyter labextension install @jupyter-widgets/jupyterlab-manager@2

Troubleshooting

Check pip packages:

pip list

Check extensions:

jupyter labextension list

Notable Dependencies

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.

Getting Started

For instructions on using TFMA, see the get started guide.

Compatible Versions

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	2.28.0	2.0.0	nightly (1.x/2.x)	0.29.0	0.29.0
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
0.26.0	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

Questions

Please direct any questions about working with TFMA to Stack Overflow using the tensorflow-model-analysis tag.