README / README.md
Zabriskije's picture
Added naming scheme for models
dd81ef7
|
raw
history blame
2.55 kB
metadata
title: Core ML Models
emoji: 🐱
pinned: false
tags:
  - coreml
  - stable-diffusion

↓↓↓ Scroll down to see models ↓↓↓

Core ML Models Repository

Thanks to Apple engineers, we can now run Stable Diffusion on Apple Silicon using Core ML!
However its hard to find compatible models and converting models isn't the easiest thing to do.
By organizing Core ML models in one place, hopefully it will be easier to find them and for everyone to benefit.

Conversion Flags

The models were converted using the following flags:
--convert-vae-decoder --convert-unet --convert-text-encoder --bundle-resources-for-swift-cli --attention-implementation {SPLIT_EINSUM or ORIGINAL}

split_einsum vs original Models

Depending on what compute unit you select, you will need to use the correct model version.
Some model versions may also generate other image sizes and are named as such.

split_einsum version is compatible with all compute unit options including Neural Engine.
original version is only compatible with CPU & GPU option.

Contributing

Before joining the community, we encourage you to have at least one model converted (that this community doesn't already have) under your account that you would be able to contribute before joining the community. This will help us to see those who can actually contribute back to the community.
Contact us on Discord if you are interested in helping out.

Models Name

Models have the following naming scheme: original model name, model version (split-einsum or original), model size (only if different from 512x512), vae name (only if different from the original vae).
Each label is separated by an underscore _, and all capitalization from the original name is preserved.
For example: stable-diffusion-2-1_original_512x768_ema-vae.

Repo Name

Repos are named with the original diffusers Hugging Face repo name prefixed by coreml-.
For example: coreml-stable-diffusion-2-1.

Repo README Contents

You can copy the beginning template for the README from one of the repo models that already exists here.
Then copy the original model's README as the body.

Repo Directory Structure

coreml-stable-diffusion-2-1
β”œβ”€β”€ README.md
β”œβ”€β”€ original
β”‚   β”œβ”€β”€ stable-diffusion-2-1_original.zip
β”‚   β”œβ”€β”€ stable-diffusion-2-1_original_512x768.zip
β”‚   └── ...
└── split_einsum
    └── stable-diffusion-2-1_split-einsum.zip