Closed ShaobinChen-AH closed 2 years ago
It seems that you are using a different type of layer norm (layer norm's parameters should be layer norm.weight/bias, not gamma/beta), can you check whether everything is consistent between train and fine-tune?
thanks a lot! I will check it.
Hello! Glad to see your excellent work! I retrained unsup simcse on my device following readme document and finetuned it on imdb dataset to improve the performance of sentiment classification. I also converted simcse to huggingface format after training the simcse. But when I load simcse when finetuning on imdb, I got the following error:
I have no idea about this. Could you point out what mistakes I made. Thanks a lot.