Hi authors,
First of all, thank you for your fantastic work and for open-sourcing the code!
I am currently studying your paper and experimenting with the repository. Could you please share the loss curves for both the training and validation sets (or perhaps provide the raw training logs)?
Seeing both curves together would be incredibly helpful for us to better understand the model's training dynamics, observe its generalization capabilities, and help with reproduction.
Thank you very much for your time and help!