Text encoder not saved in pipeline folder

[MohamedAziz-Khezami]

Hello, I am trying to train a full sdxl model, and once it finishes it gives me checkpoints folder that have unet safetensors and a pipeline folder that dosen't have the text encoder in it, so i cant convert it with the python script included. How to solve this problem please??

[bghira]

see the documentation where it mentions --save_text_encoder, this is done to save disk space as you typically already started with a copy..