-
Notifications
You must be signed in to change notification settings - Fork 6
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
Cannot reproduce the results #6
Comments
same as you ,bro, my result and parameters is listed as you can see, i wonder whether this is the problem of some parameters that are not set probably |
hi, I got the listed results with default parameters in train_generic.py. Maybe you can adjust the lr, or check if there is some error in the testing process. |
the lr is set as default 1e-5 with step optimizer, this res is my test on sha, i would try it on qnrf 3qu very much |
Did you get any progress? |
1 similar comment
Did you get any progress? |
Not yet. |
what are these values in CUR and BEST MAE and RMSE ? and what CUR stands for |
Hi, Liangzi,
Firstly, thanks for your wonderful work. I ran your code as the guideline suggested, however, I cannot get proper results.
On QNRF, I get the following result,
[2021-05-02 23:55:37,254], ## MARNet Epoch 100/100 Loss:1.257166, lr:0.0000025, [CUR]:111.3, 185.2, [Best]:95.2, 167.8
which is far away from what the paper and this repo have claimed.
On SHA, the result is
[2021-05-04 08:16:29,637], ## MARNet Epoch 300/300 Loss:1.239268, lr:0.0000002, [CUR]:68.0, 110.4, [Best]:65.9, 106.9
Could you kindly help me out
The text was updated successfully, but these errors were encountered: