search

YJiangcm / PromCSE

Code for "Improved Universal Sentence Embeddings with Prompt-based Contrastive Learning and Energy-based Learning (EMNLP 2022)"
https://arxiv.org/abs/2203.06875v2
133 stars 15 forks source link

为什么训练出来的句子相似度都是1 #14

Open Kang9779 opened 1 year ago

Kang9779 commented 1 year ago

你好，基于该仓库，用自己的领域数据训练出来的相似度都是1，是怎么回事？