Don't Copy the Teacher: Data and Model Challenges in Embodied Dialogue

TL;DR

This paper critiques current training and evaluation methods for embodied dialogue tasks, highlighting issues with imitation learning and low-level metrics, and advocates for higher-level semantic evaluation to better measure progress.

Contribution

It challenges the effectiveness of imitation learning and low-level metrics in embodied dialogue, proposing a shift towards higher-level semantic evaluation for better progress.

Findings

Models trained with imitation learning take spurious actions.

Existing models struggle to ground query utterances.

Evaluation should focus on semantic goals rather than low-level metrics.

Abstract

Embodied dialogue instruction following requires an agent to complete a complex sequence of tasks from a natural language exchange. The recent introduction of benchmarks (Padmakumar et al., 2022) raises the question of how best to train and evaluate models for this multi-turn, multi-agent, long-horizon task. This paper contributes to that conversation, by arguing that imitation learning (IL) and related low-level metrics are actually misleading and do not align with the goals of embodied dialogue research and may hinder progress. We provide empirical comparisons of metrics, analysis of three models, and make suggestions for how the field might best progress. First, we observe that models trained with IL take spurious actions during evaluation. Second, we find that existing models fail to ground query utterances, which are essential for task completion. Third, we argue evaluation should focus on higher-level semantic goals.