Is there a way to do fine tuning using STS datasets?
#27
by ijkim - opened
Hi! there.
In the paper, Used NLI format dataset to calculate the loss of positive and hard negatives for query using infoNCE loss function.
Can you tell me how to train a dataset based on the similarity score of A and B sentences in STS format, which has a different data format from the above?
Can i disable negative mode, a parameter of the infoNCE function, and proceed?