Quantized T5

[yamatazen]

How to load the quantized T5 encoder in SD.Next?

[vladmandic]

simply select in settings -> model -> text encoder before or after model load
you can also add it to quicksettings for fast access.

[yamatazen]

Does it download the encoder from Hugging Face?

[vladmandic]

yes.
if model already has t5, it will be used as-is so you don't need to download it from hf. but if you want to set explicit t5, then its downloaded.