Operationalizing Specifications, In Addition to Test Sets for Evaluating Constrained Generative Models

TL;DR

This paper advocates for enhancing the evaluation of constrained generative models by incorporating specifications alongside test sets, aiming to better assess quality amidst rapid model advancements and changing user expectations.

Contribution

It proposes using specifications as a new evaluation instrument to complement traditional test sets for more effective assessment of generative models.

Findings

Specifications improve evaluation robustness.

Evaluation methods need to adapt to model advances.

Recommendations are applicable across various tasks.

Abstract

In this work, we present some recommendations on the evaluation of state-of-the-art generative models for constrained generation tasks. The progress on generative models has been rapid in recent years. These large-scale models have had three impacts: firstly, the fluency of generation in both language and vision modalities has rendered common average-case evaluation metrics much less useful in diagnosing system errors. Secondly, the same substrate models now form the basis of a number of applications, driven both by the utility of their representations as well as phenomena such as in-context learning, which raise the abstraction level of interacting with such models. Thirdly, the user expectations around these models and their feted public releases have made the technical challenge of out of domain generalization much less excusable in practice. Subsequently, our evaluation methodologies haven't adapted to these changes. More concretely, while the associated utility and methods of interacting with generative models have expanded, a similar expansion has not been observed in their evaluation practices. In this paper, we argue that the scale of generative models could be exploited to raise the abstraction level at which evaluation itself is conducted and provide recommendations for the same. Our recommendations are based on leveraging specifications as a powerful instrument to evaluate generation quality and are readily applicable to a variety of tasks.