Discourse-Based Objectives for Fast Unsupervised Sentence Representation Learning

TL;DR

This paper introduces a new discourse-based objective for unsupervised neural sentence encoder training, enabling faster learning and improved performance by leveraging paragraph-level coherence signals.

Contribution

It proposes a discriminative objective function that significantly accelerates training of sentence encoders using discourse coherence cues.

Findings

Faster training times compared to previous methods

Models achieve strong performance in extrinsic evaluations

Utilizes paragraph-level discourse signals effectively

Abstract

This work presents a novel objective function for the unsupervised training of neural network sentence encoders. It exploits signals from paragraph-level discourse coherence to train these models to understand text. Our objective is purely discriminative, allowing us to train models many times faster than was possible under prior methods, and it yields models which perform well in extrinsic evaluations.