CoT-BERT: Enhancing Unsupervised Sentence Representation through Chain-of-Thought

TL;DR

CoT-BERT introduces a novel unsupervised sentence representation method that leverages Chain-of-Thought reasoning, an advanced contrastive loss, and template denoising to improve embedding quality without external components.

Contribution

This paper proposes CoT-BERT, a new approach that enhances unsupervised sentence embeddings by integrating Chain-of-Thought reasoning with improved contrastive learning and denoising strategies, avoiding external modules.

Findings

Outperforms established baselines in unsupervised sentence representation tasks.

Uses only pre-trained models without external components.

Demonstrates significant improvements through rigorous experiments.

Abstract

Unsupervised sentence representation learning aims to transform input sentences into fixed-length vectors enriched with intricate semantic information while obviating the reliance on labeled data. Recent strides within this domain have been significantly propelled by breakthroughs in contrastive learning and prompt engineering. Despite these advancements, the field has reached a plateau, leading some researchers to incorporate external components to enhance the quality of sentence embeddings. Such integration, though beneficial, complicates solutions and inflates demands for computational resources. In response to these challenges, this paper presents CoT-BERT, an innovative method that harnesses the progressive thinking of Chain-of-Thought reasoning to tap into the latent potential of pre-trained models like BERT. Additionally, we develop an advanced contrastive learning loss function and propose a novel template denoising strategy. Rigorous experimentation demonstrates that CoT-BERT surpasses a range of well-established baselines by relying exclusively on the intrinsic strengths of pre-trained models.