esmjax icon indicating copy to clipboard operation
esmjax copied to clipboard
verified publisher icon irhum

Have you managed to train the model

Open sarthak-314 opened this issue 2 years ago • 0 comments

The training time for the model is very slow - 400 hours per epoch for 3B parameter as compared to 1.5 hours per epoch for my Keras data-parallel implementation of 650M model

sarthak-314 avatar May 13 '23 20:05 sarthak-314