Hello, I have dataset with multiple summaries written by independent human annotatos, I want to know what is the standard approach in training? Is it simply duplicating the input for each summary or concatenate all the summaries in the target?
Thank you