How to train the BebyGAN network using upscaling factors 2 and 3 on different dataset?

[Techzist16]

I have tried to run the BebyGAN network on my own dataset for upsacling factor 2 and 3. To do this, I must first train the RRDB network with l1 loss. But I'm not sure where I should adjust the code structure to make it trainable. Please provide a guideline for simple comprehension.
@fenglinglwb

[fenglinglwb]

Hi, thanks for your attention. Just modify the upsampling module as you wish.

[Techzist16]

Thanks for your prompt answer. However, I attempted to run the RRDB network. Because my dataset is different, it is not appropriately fitted to the model. Can you tell me how to train the RRDB network on different datasets?
@fenglinglwb

[fenglinglwb]

I didn't get your point. Do you mean how to adopt a new dataset (not DIV2K or Flickr2K) to train RRDB? If yes, you may need to manually prepare low-resolution images and provide the location info in dataset/init.py.

[Techzist16]

Okay. Is it possible for me to get your email address so that I can explain my concern further?
@fenglinglwb

[fenglinglwb]

Sure. You can contact me at wenboli@cse.cuhk.edu.hk.