Where to specify "train text encoder"?

[rexroth0619]

Noticed in original Diffusers colab there's a section saying "training text encoder has a much better result." Just wondering where should I enable text encoder training with current implementation? Thanks