arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2211.01587cs.CL
收藏

从大型预训练模型中获取知识,用于无监督的基于知识的对话

Eliciting Knowledge from Large Pre-Trained Models for Unsupervised Knowledge-Grounded Conversation

The Chinese University of Hong Kong(香港中文大学)

Yanyang Li, Jianqiao Zhao, Michael R. Lyu, Liwei Wang

摘要:

大规模预训练的最新进展为大型模型提供了从原始文本中学习知识的潜力。因此,很自然地会问,是否有可能利用这些大型模型作为下游任务的知识库。在这项工作中,我们在无监督的知识基础对话中回答了上述问题。我们探索各种方法,最好地从大型模型中获取知识。我们的人类研究表明,尽管存在幻觉,但大型模型具有独特的优势,能够输出常识并总结无法从搜索引擎直接检索的事实。为了更好地利用这些产生的知识在对话生成,我们把产生的知识作为一个嘈杂的知识源,并提出了基于后验的重新加权以及嘈杂的训练策略。两个基准的实证结果显示出优于国家的最先进的方法。

英文摘要:

Recent advances in large-scale pre-training provide large models with the potential to learn knowledge from the raw text. It is thus natural to ask whether it is possible to leverage these large models as knowledge bases for downstream tasks. In this work, we answer the aforementioned question in unsupervised knowledge-grounded conversation. We explore various methods that best elicit knowledge from large models. Our human study indicates that, though hallucinations exist, large models post the unique advantage of being able to output common sense and summarize facts that cannot be directly retrieved from the search engine. To better exploit such generated knowledge in dialogue generation, we treat the generated knowledge as a noisy knowledge source and propose the posterior-based reweighing as well as the noisy training strategy. Empirical results on two benchmarks show advantages over the state-of-the-art methods.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑