disco: a toolkit for Distributional Control of Generative Models
This work provides a practical solution for researchers and practitioners to mitigate biases and control outputs in generative models, though it is incremental as it focuses on tooling rather than new algorithmic breakthroughs.
The paper introduces disco, an open-source Python toolkit designed to address the limitations of generative models, such as reproducing undesirable biases and overlooking important patterns, by enabling distributional control of features in model outputs, making these techniques more accessible to the broader public.
Pre-trained language models and other generative models have revolutionized NLP and beyond. However, these models tend to reproduce undesirable biases present in their training data. Also, they may overlook patterns that are important but challenging to capture. To address these limitations, researchers have introduced distributional control techniques. These techniques, not limited to language, allow controlling the prevalence (i.e., expectations) of any features of interest in the model's outputs. Despite their potential, the widespread adoption of these techniques has been hindered by the difficulty in adapting complex, disconnected code. Here, we present disco, an open-source Python library that brings these techniques to the broader public.