refine_mode hyperstyle?

[artificialnouveau]

I see in the readme there is a possibility of doing the refinement with hyperstyle:

This uniform inversion process can easily combine two methods. Users can try any combination of methods, not limited to those employed by the original authors. For example, GANInverter makes it possible to connect ReStyle with HyperStyle by --embed_mode restyle --refine_mode hyperstyle or PTI + e4e by --embed_mode e4e --refine_mode pti.

However, in the inference/two_stage_inference.py, I do not see a option to have hyperstyle as an input for the refine_mode. Is there a script for this? If not, how could I implement it?

[caopulan]

We update the repo and hyperstyle is supported. However, the updated codes were implemented one year ago and we are not working on inversion now, there might be some errors.

[artificialnouveau]

Thank you for your fast reply @caopulan, so where can I find the updated code with the hyperstyle support? I can try to adapt it.