Is there someone who can share implemented script for --lock-text-unlocked-layers for "Native Clip Model"?

[Vikho]

I want to freeze both of the image and text tower for fine-tuning, but now the repo is not support for non-hf text transfomer.