3 ms·
Transformers require a lot of data to converge. There’s a reason tree models are still king of kaggle even though transformers have been around for 5 years now.
by janalsncm 4y ago
Transformers require a lot of data to converge. There’s a reason tree models are still king of kaggle even though transformers have been around for 5 years now.