Hi, Dat Tran! What you did is amazing! I learned a lot ! Thank you !
Jianlang Huang

The easiest way to prevent overfitting is to add more data. Also although I trained it more than 22k I used an earlier checkpoint that was created. You should stop training if the loss doesn’t improve which also can help.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.