Hello @thunil @tom-doerr & all,
I would like to fine-tune the available pre-trained model with my own video frames. However, when I start the training (after having set the variable "old_model" with the proper path), the printed message is the following:
Prepare to load 100 weights from the pre-trained model for generator and fnet
Prepare to load 0 weights from the pre-trained model for discriminator
I suppose that the pre-trained model available for inference reasonably contains only the weights of the generator, which are the ones that are useful for the production of super-resolved frames. However, I would need also the weights of the discriminator to fine-tune the GAN (please, correct me if I am wrong, I do not have much experience with GANs).
Is it possible to access the pre-trained discriminator?
Hello @thunil @tom-doerr & all, I would like to fine-tune the available pre-trained model with my own video frames. However, when I start the training (after having set the variable "old_model" with the proper path), the printed message is the following:
Prepare to load 100 weights from the pre-trained model for generator and fnet
Prepare to load 0 weights from the pre-trained model for discriminator
I suppose that the pre-trained model available for inference reasonably contains only the weights of the generator, which are the ones that are useful for the production of super-resolved frames. However, I would need also the weights of the discriminator to fine-tune the GAN (please, correct me if I am wrong, I do not have much experience with GANs). Is it possible to access the pre-trained discriminator?