ReadOnce Transformers: Reusable Representations of Text for Transformers

Abstract

We present ReadOnce Transformers, an approach to convert a transformer-based model into one that can build an information-capturing, task-independent, and compressed representation of text. The resulting representation is reusable across different examples and tasks, thereby requiring a document shared across many examples or tasks to only be \emph{read once}. This leads to faster training and evaluation of models. Additionally, we extend standard text-to-text transformer models to Representation+Text-to-text models, and evaluate on multiple downstream tasks: multi-hop QA, abstractive QA, and long-document summarization. Our one-time computed representation results in a 2x-5x speedup compared to standard text-to-text models, while the compression also allows existing language models to handle longer documents without the need for designing new pre-trained models.

arXiv commentAccepted to ACL 2021(Camera Ready)

References in corpus (9)

• Distilling the Knowledge in a Neural Network
• Longformer: The Long-Document Transformer
• Domain Adaptation for Statistical Classifiers
• Skip-Thought Vectors
• Big Bird: Transformers for Longer Sequences
• Unsupervised Question Answering by Cloze Translation
• Pre-trained Summarization Distillation
• DeFormer: Decomposing Pre-trained Transformers for Faster Question Answering
• Contextual Lensing of Universal Sentence Representations