Hi everyone,
I want to try and fine-tune gpt-2 to fix and grade essays that are about 250 - 300 words long. Very simple and also similarly structured essays that are mostly for foreign language learners. I understand that I should feed fixed and unfixed essays to the model as well as incorporate some grammar training, but I am not sure how to start with feeding fixed and unfixed essays because the way I understand it is that you feed it a big set of data which it will sample from. How exactly would this work with the fixed-unfixed set?
I think that I should be looking at Seq2Seq but I also fail to find many resources for that.
If someone could point me to a tutorial or show me where to start that would be really appreciated.