CoT-BERT: Enhancing Unsupervised Sentence Representation through Chain-of-Thought
This work addresses the challenge of improving sentence embeddings without labeled data for natural language processing applications, though it appears incremental by building on existing contrastive learning and prompt engineering methods.
The paper tackles the problem of plateauing performance in unsupervised sentence representation learning by introducing CoT-BERT, which uses Chain-of-Thought reasoning to enhance pre-trained models like BERT, resulting in superior performance over established baselines.
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.