DialoGLUE: A Natural Language Understanding Benchmark for Task-Oriented Dialogue

TL;DR

DialoGLUE is a comprehensive benchmark for evaluating natural language understanding in task-oriented dialogue, promoting research in domain adaptation, transfer learning, and sample efficiency.

Contribution

It introduces a new benchmark with diverse datasets and baseline models, advancing the development of more adaptable dialogue systems.

Findings

Baseline models outperform vanilla BERT on 5 out of 7 tasks.

Pre-training on dialogue data improves task performance.

Benchmark facilitates progress in general task-oriented dialogue modeling.

Abstract

A long-standing goal of task-oriented dialogue research is the ability to flexibly adapt dialogue models to new domains. To progress research in this direction, we introduce DialoGLUE (Dialogue Language Understanding Evaluation), a public benchmark consisting of 7 task-oriented dialogue datasets covering 4 distinct natural language understanding tasks, designed to encourage dialogue research in representation-based transfer, domain adaptation, and sample-efficient task learning. We release several strong baseline models, demonstrating performance improvements over a vanilla BERT architecture and state-of-the-art results on 5 out of 7 tasks, by pre-training on a large open-domain dialogue corpus and task-adaptive self-supervised training. Through the DialoGLUE benchmark, the baseline methods, and our evaluation scripts, we hope to facilitate progress towards the goal of developing more general task-oriented dialogue models.