What dose "inverting 3 out of 10 templates" means?

[Cola-any]

Hi, I am new in this field and I am very interested in your excellent research.

In your paper, you mentioned

To compare the effect of template diversity on model performance, we evaluate three scenarios with different diversities: using a single template (single), employing 10 templates for each dataset (multi), and inverting 3 out of 10 templates (multi+flip).

It's hard for me to understand what "inverting 3 out of 10 templates (multi+flip)" means. I would really appreciate it if you would explain this for me.

Sincerely.