Has anyone managed to use the code in examples/seq2seq on the newsroom dataset used in the pegasus paper?
If you have good data (one line per example for source and target), but no code, I would find a google drive/s3 link to the data useful! Thanks!