Ming Zhu

Can you please share the untokenized version of [transcoder_test_set.zip](https://dl.fbaipublicfiles.com/transcoder/test_set/transcoder_test_set.zip)? I would like to tokenize it in a different way. Thank you in advance!

I see the data are preprocessed into multi-hot vectors in matlab. But I wonder how is it preprocessed? If I want to switch the dataset, what should I do to...