Always encounter "AssertionError: You do not have CLIP state dict!"

[shiboyang]

Use some of ckpt to generate the image will raise "You do not have CLIP state dict!"
Has anyone encounted this situation?
the sdxl-base model can work, however the sdxl-refine model sometime not work

[Jonseed]

I'm also seeing this error on the SDXL refiner. Not sure how to fix it.

[dallas4u]

Make sure your text encoders (clip_l and t5-xxl) are in the \models\text_encoder folder
Make sure your VAE are in the \models\VAE folder
Then, start forge and test

[Jonseed]

yes, they are in the correct locations, and I still get the error.

[usigg3]

At least with Flux you need to select them in the "VAE / Text Encoder" window, you can select several there.