Adapting Pretrained Text-to-Text Models for Long Text Sequences
Wenhan Xiong, Anchit Gupta, Shubham Toshniwal, Yashar Mehdad, Scott Yih · 2023
We present an empirical study of adapting an existing pretrained text-to-text model for long-sequence inputs.Through a comprehensive study along three axes of the pretraining pipeline -model architecture, optimization objective, and pretraining corpus, we propose an effective recipe to build long-context models from existing short-context models.Specifically, we replace the full attention in transformers with pooling-augmented blockwise attention, and pretrain the model with a maskedspan prediction task with spans of varying lengths.In terms of the pretraining corpus, we find that using randomly concatenated shortdocuments from a large open-domain corpus results in better performance than using existing long document corpora, which are typically limited in their domain coverage.With these findings, we build a long-context model that achieves competitive performance on long-text QA tasks and establishes the new state of the art on five long-text summarization datasets, often outperforming previous methods with larger model sizes.