I have a task which involves Machine Reading Comprehension (Question Answering). I was thinking of using the same approach for LM, but the data is quite different. There are multiple answers(10) for each question. We are required to pick the best one. There hasn’t been much implementation available which uses transfer learning (they all use embeddings). What approaches can be used with the current method (covered in lesson 4)?
There doesn’t seem to be much interest in this topic for some reason. Is there anyone who managed to create a MRC model with fastai?