-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Why HR and MRR become smaller after each epoch? #1
Comments
Sorry, I am not the author of the original paper. I tried to reproduce the paper but found it difficult to achieve the effect in the paper. I have fixed the code。 |
@slientGe Thank you,it works now. |
hello,did it work now? I ran the code and the results are the same like you,which is becoming samller and smaller. |
Hello,did you fix the code? I ran the code but the results of HR@50 and MRR became smaller after each epoch. |
In fact, you don’t need to run so many epoches each time, small datasets are easy to overfit, resulting in no improvement in results |
Thank you,I understand.I found the rusults were becoming larger before the 6 epoch,after that,which was smaller.It's overfit. |
epoch 1, mean_loss0.130139, test HR@50: 0.322375, test MRR: 0.0353417
epoch 2, mean_loss0.0528188, test HR@50: 0.336161, test MRR: 0.0330826
epoch 3, mean_loss0.0253366, test HR@50: 0.316013, test MRR: 0.0339782
epoch 4, mean_loss0.0166349, test HR@50: 0.306469, test MRR: 0.0332131
………………
epoch 28, mean_loss0.00114451, test HR@50: 0.246023, test MRR: 0.0233681
epoch 29, mean_loss0.00107539, test HR@50: 0.244963, test MRR: 0.0247977
epoch 30, mean_loss0.00104046, test HR@50: 0.240721, test MRR: 0.0242149
The paper states that these two values should be larger as possible.I ran the original author's code and it got smaller and smaller, I didn't understand why.
Could you please tell me why it is calculated like this?
The text was updated successfully, but these errors were encountered: