Improving Prediction Backward-Compatiblility in NLP Model Upgrade with Gated Fusion

TL;DR

This paper introduces Gated Fusion, a novel method to improve backward compatibility in NLP model upgrades, significantly reducing regression errors and facilitating smoother transitions between model versions.

Contribution

The paper proposes Gated Fusion, a new approach that learns to combine old and new model predictions to minimize regression errors during upgrades.

Findings

Reduces regression errors by 62% on average

Outperforms baseline by 25% in regression reduction

Effective across multiple model upgrade scenarios

Abstract

When upgrading neural models to a newer version, new errors that were not encountered in the legacy version can be introduced, known as regression errors. This inconsistent behavior during model upgrade often outweighs the benefits of accuracy gain and hinders the adoption of new models. To mitigate regression errors from model upgrade, distillation and ensemble have proven to be viable solutions without significant compromise in performance. Despite the progress, these approaches attained an incremental reduction in regression which is still far from achieving backward-compatible model upgrade. In this work, we propose a novel method, Gated Fusion, that promotes backward compatibility via learning to mix predictions between old and new models. Empirical results on two distinct model upgrade scenarios show that our method reduces the number of regression errors by 62% on average, outperforming the strongest baseline by an average of 25%.