Hugging Face's logo

coreml-community
/
coreml-YamersRealistic-v4_SDXL_8-bit

Text-to-Image
Core ML
stable-diffusion
Not-For-All-Audiences
Model card Files Community
1

missing converted models?

#1
by sa902 - opened
Discussion
sa902

Seems like the repo is missing converted models and only has originals?
Can you post converted models?

zhuguanyu
 Core ML Models org

The originals are the converted modals. "Original" is a type of converted modal, while the other type is called "SPLIT_EINSUM".

sa902

Okay, thank you for the clarification! I assumed that all models were supposed to run on the ANE but I realise now its only the SPLIT_EINSUM models. Bit of a skill issue from me!

deleted

Original = faster but uses more RAM
Split_Einsum = more efficient but much longer wait times

If you have a macbook with 16gb or more, you should be fine to use Original models. If you have any other questions don't hesitate to ask! It's a bit of a steep learning curve at first.

Sign up or log in to comment