Sequence Modeling with Unconstrained Generation Order

Abstract

The dominant approach to sequence generation is to produce a sequence in some predefined order, e.g. left to right. In contrast, we propose a more general model that can generate the output sequence by inserting tokens in any arbitrary order. Our model learns decoding order as a result of its training procedure. Our experiments show that this model is superior to fixed order models on a number of sequence generation tasks, such as Machine Translation, Image-to-LaTeX and Image Captioning.

Camera-ready version for NeurIPS2019

Cited by in corpus (5)

• PAM: Understanding Product Images in Cross Product Category Attribute Extraction
• Neural Machine Translation: Challenges, Progress and Future
• Fast Interleaved Bidirectional Sequence Generation
• Syntax-driven Iterative Expansion Language Models for Controllable Text Generation
• Cue-word Driven Neural Response Generation with a Shrinking Vocabulary