You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When Kaggle trains the algorithm it only goes through 5 epochs due to earlystopping. Unfortunately, this means that it's accuracy is often sub-70 on SwDa datasets. Is it possible for kaggle to create longer checkpoints?