search

kongds / scaling_sentemb

Scaling Sentence Embeddings with Large Language Models
85 stars 4 forks source link

Maybe errors in pre-print paper #3

Closed YuboFeng2023 closed 10 months ago

YuboFeng2023 commented 10 months ago

Dear author,

What a fabulous sentence embedding model you have created! But there are show error(s) that need to be corrected in pre-print paper:

  1. Temperature of contrastive learning. In your paper, the temperature is claimed to be 0.5, but I found it should be 0.05 according to your code. All in all, which one is correct?

Best wishes!

kongds commented 10 months ago

Thank you for your advice. The temperature of contrastive learning should be 0.05. I will correct this error in the next version of the paper.