reCSE: Portable Reshaping Features for Sentence Embedding in Self-supervised Contrastive Learning

TL;DR

reCSE introduces a novel feature reshaping method for sentence embeddings in self-supervised contrastive learning, improving efficiency and performance in semantic similarity tasks.

Contribution

It presents a new feature reshaping approach that enhances sentence representation quality and universality across contrastive learning frameworks.

Findings

Achieved competitive results in semantic similarity tasks.

Reduced GPU memory consumption compared to existing models.

Proven universal applicability of feature reshaping method.

Abstract

We propose reCSE, a self supervised contrastive learning sentence representation framework based on feature reshaping. This framework is different from the current advanced models that use discrete data augmentation methods, but instead reshapes the input features of the original sentence, aggregates the global information of each token in the sentence, and alleviates the common problems of representation polarity and GPU memory consumption linear increase in current advanced models. In addition, our reCSE has achieved competitive performance in semantic similarity tasks. And the experiment proves that our proposed feature reshaping method has strong universality, which can be transplanted to other self supervised contrastive learning frameworks and enhance their representation ability, even achieving state-of-the-art performance. Our code is available at https://github.com/heavenhellchen/reCSE.