Oct 2, 2023
Hi Enzodefi!
Flan-T5 is based on exactly the same codebase as mT5; the only difference is how these models were pretrained.
Thus, all my code for vocabulary manipulation is expected to work with Flan-T5.
Hi Enzodefi!
Flan-T5 is based on exactly the same codebase as mT5; the only difference is how these models were pretrained.
Thus, all my code for vocabulary manipulation is expected to work with Flan-T5.
NLP researcher at FAIR, Meta. Low-resource language enthusiast. See daviddale.ru.