Finetune

#2
by JSLuo888 - opened

Hi,

I am interested to know if you have a workflow for fine-tuning the model, and wonder if you have any experience on how many samples would be sufficient for finetuning. Thanks!

siqi-liu-paige changed discussion status to closed

We do not finetune the model for our internal R&D purpose. It's always used as a frozen embedding generation model so far. However, keen to learn if you found anything interesting by finetuning it it.

Sign up or log in to comment