Ditto: A Simple and Efficient Approach to Improve Sentence Embeddings
Abstract
Prior studies diagnose the anisotropy problem in sentence representations from pre-trained language models, e.g., BERT, without fine-tuning. Our analysis reveals that the sentence embeddings from BERT suffer from a bias towards uninformative words, limiting the performance in semantic textual similarity (STS) tasks. To address this bias, we propose a simple and efficient unsupervised approach, Diagonal Attention Pooling (Ditto), which weights words with model-based importance estimations and computes the weighted average of word representations from pre-trained models as sentence embeddings. Ditto can be easily applied to any pre-trained language model as a postprocessing operation. Compared to prior sentence embedding approaches, Ditto does not add parameters nor requires any learning. Empirical evaluations demonstrate that our proposed Ditto can alleviate the anisotropy problem and improve various pre-trained models on STS tasks.
Keywords
Cite
@article{arxiv.2305.10786,
title = {Ditto: A Simple and Efficient Approach to Improve Sentence Embeddings},
author = {Qian Chen and Wen Wang and Qinglin Zhang and Siqi Zheng and Chong Deng and Hai Yu and Jiaqing Liu and Yukun Ma and Chong Zhang},
journal= {arXiv preprint arXiv:2305.10786},
year = {2023}
}
Comments
8 pages, accepted by EMNLP 2023 short paper, the source code can be found at https://github.com/alibaba-damo-academy/SpokenNLP/tree/main/ditto