How to train text encoder for dreambooth inpaint lora?

[XYZ-916]

There is train_text_encoder in utils/train_dreambooth_lora_inpaint.py. But it seems not work.

Thank you!