Open JLA0809 opened 4 days ago
I think there could be an indendation error in the main training loop in the utils/utils.py in the original main branch. In my opinion the optimizer has to be called for every batch, in order to use the information of the whole training dataset.
I think there could be an indendation error in the main training loop in the utils/utils.py in the original main branch. In my opinion the optimizer has to be called for every batch, in order to use the information of the whole training dataset.