The training time is very large #1
Description
Hello,
I ran your code for a dataset of 2300 URLs and encountered an issue where the training time is very large, taking 32 hours for 50 epochs. On average, each epoch takes about 40 minutes.
I encountered an "unnecessary retracing" error during the data loading, but I think that's not a major issue. After that, the training time for each epoch is as shown in the image.
I hope you can explain to me why the training time is so long, or if I made a mistake at any step.
Thank you very much!
