search

arneschneuing / DiffSBDD

A Euclidean diffusion model for structure-based drug design.
MIT License
339 stars 74 forks source link

Why used different methods to calculate loss-t for training and validation? #44

Open Yuhan-Fei opened 4 months ago

Yuhan-Fei commented 4 months ago

Thank you for the great works!

I wondering why you used different loss_t for training and validation or Is there any problem to use the same method to calculate the loss_t?

For training: loss_t = 0.5 * (error_t_lig + error_t_pocket)

For validaiton: loss_t = -self.T 0.5 SNR_weight * (error_t_lig + error_t_pocket)

I am looking forward your reply.

Best