Closed nikifori closed 9 months ago
@nikifori
Please accept my apologies for the late reply. I've been very busy lately because of the ICML submission.
To answer your question: Take a look on : https://github.com/PiotrNawrot/nanoT5/issues/11 and https://github.com/PiotrNawrot/nanoT5/issues/21
Training on a different dataset seems like a common question so I'll try to add some more information about it to the README.
I'm closing this Issue for now, please feel free to re-open if you have any other questions.
Is there any option, to pre-train T5-based models on different dataset than C4 in a self-supervised manner?