May I ask to which range will the converged loss value fall when training the representation network? (e.g., based on parameters in your paper, the ultimately converged loss value is less than 5, or less than 1, or 0.1 ? ) This can be helpful during training.
Hi, thank you for this impressive work!
May I ask to which range will the converged loss value fall when training the representation network? (e.g., based on parameters in your paper, the ultimately converged loss value is less than 5, or less than 1, or 0.1 ? ) This can be helpful during training.
Thank you a lot!