academic-budget-bert icon indicating copy to clipboard operation
academic-budget-bert copied to clipboard

Repository containing code for "How to Train BERT with an Academic Budget" paper

Results 11 academic-budget-bert issues
Sort by recently updated
recently updated
newest added

In the first epoch of pretraining, grad overflow happened in every iteration. Also, the evaluation loss of some epochs is null, after about the 17th epoch. It looks like the...