2
$\begingroup$

I am trying to reproduce the results of paper that fine-tunes T5 (a deep learning language model) on a dataset. In, the paper they say they fine-tune all their models for 10k steps. For the base-version of T5, they use a batch size of 128, for the large version of T5 they use a batch size of 32, and for the XL version of T5 they use a batch size of 16.

Does this mean they are fine tuning the base version for 8 times more epochs than the XL version of T5?

$\endgroup$

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.