Hello,

Mist is a fascinating project, and we are interested in following its experimental settings for textual inversion. However, we have encountered an issue with the LSUN-CAT dataset. It appears that the images of cats in this dataset are not consistent; they may not all represent the same cat.

Given this situation, we are unsure about how to proceed with textual inversion. Could you please provide some guidance on how to select cats for textual inversion from the LSUN-CAT dataset? Is it recommended to choose cats randomly and form groups for the inversion process?

Any advice or insights you can provide on this matter would be highly appreciated.

Thank you!

Best regards.