ChatZero:Zero-shot Cross-Lingual Dialogue Generation via Pseudo-Target Language

TL;DR

ChatZero is a novel zero-shot cross-lingual dialogue generation model that leverages code-switching and contrastive learning to effectively generate dialogues in low-resource languages without requiring large-scale dialogue data.

Contribution

The paper introduces ChatZero, a new end-to-end model that uses cross-lingual code-switching and unsupervised contrastive learning for zero-shot multilingual dialogue generation.

Findings

Achieves over 90% of supervised performance in zero-shot setting.

Outperforms existing baselines on multilingual dialogue datasets.

Demonstrates state-of-the-art results in low-resource language scenarios.

Abstract

Although large language models(LLMs) show amazing capabilities, among various exciting applications discovered for LLMs fall short in other low-resource languages. Besides, most existing methods depend on large-scale dialogue corpora and thus building systems for dialogue generation in a zero-shot scenario remains a considerable challenge. To address this challenge, we propose a novel end-to-end zero-shot dialogue generation model ChatZero based on cross-lingual code-switching method. First, we construct code-switching language and pseudo-target language with placeholders. Then for cross-lingual semantic transfer, we employ unsupervised contrastive learning to minimize the semantics gap of the source language, code-switching language, and pseudo-target language that are mutually positive examples in the high dimensional semantic space. Experiments on the multilingual DailyDialog and DSTC7-AVSD datasets demonstrate that ChatZero can achieve more than 90\% of the original performance under the zero-shot case compared to supervised learning, and achieve state-of-the-art performance compared with other baselines.