DocPrompt: Large-scale continue pretrain for zero-shot and few-shot document question answering

TL;DR

This paper introduces DocPrompt, a comprehensive approach combining novel data generation, multi-stage training, and ensemble methods to significantly enhance zero-shot and few-shot document question answering performance.

Contribution

The paper presents a new weakly supervised data generation technique, multi-stage training process, and ensemble approach, achieving state-of-the-art results in document question answering.

Findings

Achieved state-of-the-art results on 4 document QA tasks.

Significantly reduced annotation and labor costs.

Enhanced zero-shot and few-shot performance.

Abstract

In this paper, we propose Docprompt for document question answering tasks with powerful zero-shot and few-shot performance. We proposed a novel weakly supervised data generation method, a novel multl-stage training method and a novel understanding model \& generation model ensemble method. We achieved state-of-the-art performance on 4 document question answering tasks. This method greatly improves the delivery efficiency and model performance of document question answering customer projects, reducing annotation costs and labor costs. Our demo can be found at https://huggingface.co/spaces/PaddlePaddle/ERNIE-Layout.