Comparison with sentence-transformers

[MastafaF]

Hi, any comparison with sentence-transformers form https://github.com/UKPLab/sentence-transformers ?

That would be really interesting indeed. 😃

[BinWang28]

Hi,
we do have the comparison and also used the pre-trained models from UKPLab.
More details: https://arxiv.org/abs/2002.06652

[MastafaF]

Really interesting paper. Good job Bin!
I will try to compare pre-trained models from UKPLab and your pretrained embeddings on WMT similarity task. From the README, I assume that I should use sen_emb.py to get sentence embeddings from a given string, would you agree?

[BinWang28]

Yep, sure.

[MastafaF]

Hi @BinWang28 , how would you go to provide sentence-embeddings from your code?
I checked sen_emb.py after looking at example2.sh, but I cannot see where is your pretrained model exactly? A snippet of code like sentence-transformers does would be very useful. 😀

[BinWang28]

Hi, We have uploaded the pre-trained model on the server. It will automatically be downloading when you run the script.

[MastafaF]

Hi @BinWang28 , are you planning on sharing multilingual models as well? That would be of interest for me.