TensorFlow Model Remediation is a library that provides solutions for machine learning practitioners working to create and train models in a way that reduces or eliminates user harm resulting from underlying performance biases.

You can install the package from pip:

$ pip install tensorflow-model-remediation

Note: Make sure you are using TensorFlow 2.x.

This library will ultimately contain a collection of techniques for addressing a wide range of concerns. For now it contains a single technique, MinDiff, which can help reduce performance gaps between example subgroups.

We recommend starting with the overview guide or trying it interactively in our tutorial notebook.

from tensorflow_model_remediation import min_diff
import tensorflow as tf

# Start by defining a Keras model.
original_model = ...

# Set the MinDiff weight and choose a loss.
min_diff_loss = min_diff.losses.MMDLoss()
min_diff_weight = 1.0  # Hyperparamater to be tuned.

# Create a MinDiff model.
min_diff_model = min_diff.keras.MinDiffModel(
    original_model, min_diff_loss, min_diff_weight)

# Compile the MinDiff model as you normally would do with the original model.
min_diff_model.compile(...)

# Create a MinDiff Dataset and train the min_diff_model on it.
min_diff_model.fit(min_diff_dataset, ...)

If you're interested in learning more about responsible AI practices, including fairness, please see Google AI's Responsible AI Practices.

tensorflow/model_remediation is Apache 2.0 licensed. See the LICENSE file.