search

4AI / BeLLM

Code for BeLLM: Backward Dependency Enhanced Large Language Model for Sentence Embeddings (NAACL2024)
https://arxiv.org/abs/2311.05296
MIT License
5 stars 0 forks source link

Call for Code Release #1

Closed AnonymXXXXX closed 6 months ago

AnonymXXXXX commented 6 months ago

First of all, congratulations on your paper being accepted to NAACL 2024.

A few months ago, I read the preprint you uploaded to ArXiv, where your model, based on LLaMA2, achieved a Spearman correlation coefficient score of 86.01 across seven STS benchmark tasks.

Recently, I noticed that you've updated your paper and that the corresponding experimental results have improved from 86.01 to 86.26.

We have been unable to replicate these results and are therefore eager to access the corresponding code. As per the ReadMe file, it was mentioned that the code would be made available by the end of March 2024. Given that it's now April 5,2024, and the code has not yet been released, could you please share the code or provide an update on its availability?

Thank you for your attention to this matter. We are keen to explore and build upon your work.

SeanLee97 commented 6 months ago

@AnonymXXXXX Thank you for following our work and sorry for the delay. It should be released last week. But I was busy with the ACL rebuttal. Thus the schedule changed. I will return to work on it next week. Will let you know when push the code and pretrained weights.

AnonymXXXXX commented 6 months ago

Thanks for your reply!

SeanLee97 commented 6 months ago

hi, I have pushed the code. More details are incoming

AnonymXXXXX commented 6 months ago

Thank you very much for open source your code, we will study it carefully.