Plug-and-Play Recipe Generation with Content Planning

TL;DR

This paper introduces a novel framework for multi-sentence recipe generation that explicitly models global content planning, achieving state-of-the-art results through a plug-and-play approach on the Recipe1M+ benchmark.

Contribution

It presents a low-cost, effective method for global content planning in text generation, addressing limitations of existing controlled generation techniques.

Findings

Achieves state-of-the-art performance on recipe generation.

Outperforms existing methods in automatic and human evaluations.

Effectively models global content plan in a plug-and-play manner.

Abstract

Recent pre-trained language models have shown promising capabilities in generating fluent and realistic natural language text. However, generating multi-sentence text with global content planning has been a long-existing research question. Current approaches for controlled text generation can hardly address this issue, as they usually condition on single known control attributes. In this study, we propose a low-cost yet effective framework which explicitly models the global content plan of the generated text. Specifically, it optimizes the joint distribution of the natural language sequence and the global content plan in a plug-and-play manner. We conduct extensive experiments on the well-established Recipe1M+ benchmark. Both automatic and human evaluations verify that our model achieves the state-of-the-art performance on the task of recipe generation