ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation

TL;DR

ERNIE-GEN introduces a multi-flow pre-training framework for natural language generation that reduces exposure bias and improves performance across various tasks with less data and fewer parameters.

Contribution

It proposes a novel span-by-span generation and multi-granularity sampling approach to enhance pre-training effectiveness and alignment between training and inference.

Findings

Achieves state-of-the-art results on multiple language generation tasks.

Uses fewer pre-training data and parameters than existing methods.

Demonstrates improved generation quality closer to human writing.

Abstract

Current pre-training works in natural language generation pay little attention to the problem of exposure bias on downstream tasks. To address this issue, we propose an enhanced multi-flow sequence to sequence pre-training and fine-tuning framework named ERNIE-GEN, which bridges the discrepancy between training and inference with an infilling generation mechanism and a noise-aware generation method. To make generation closer to human writing patterns, this framework introduces a span-by-span generation flow that trains the model to predict semantically-complete spans consecutively rather than predicting word by word. Unlike existing pre-training methods, ERNIE-GEN incorporates multi-granularity target sampling to construct pre-training data, which enhances the correlation between encoder and decoder. Experimental results demonstrate that ERNIE-GEN achieves state-of-the-art results with a much smaller amount of pre-training data and parameters on a range of language generation tasks, including abstractive summarization (Gigaword and CNN/DailyMail), question generation (SQuAD), dialogue generation (Persona-Chat) and generative question answering (CoQA).