---
license: creativeml-openrail-m
tags:
- text-to-image
- stable-diffusion
- stable-diffusion-1.5
- dreamshaper
- coreml
- fp16
- ios
base_model:
- Lykon/dreamshaper-8
pipeline_tag: text-to-image
library_name: ml-stable-diffusion
---

# DreamShaper 8 — Core ML FP16 for LocalMuse

This repository contains an unquantized FP16 Core ML format conversion of
[DreamShaper 8 by Lykon](https://huggingface.co/Lykon/dreamshaper-8) for
on-device inference in the LocalMuse iOS app.

## Provenance

- Source repository: `Lykon/dreamshaper-8`
- Pinned source revision: `a7e52b98680b1ba8ff7bce97c7f9f2e2e5337917`
- Architecture: Stable Diffusion 1.5
- Fixed resolution: 512×512
- Storage precision: FP16
- Quantization/palettization: none
- UNet attention graph: `SPLIT_EINSUM_V2`
- Core ML deployment target: iOS 17
- Core ML execution policy in LocalMuse: CPU + GPU
- Conversion tooling: Apple `ml-stable-diffusion` commit
  `e12202c1f6405b83918b58a5d097cd61e3e1f702`, Core ML Tools 8.3.0

The UNet is split into two compiled graphs so each weight file remains below
1 GB. Splitting changes only the graph packaging and does not quantize the
weights. Text encoder, VAE decoder and VAE encoder are included, so both
text-to-image and image-to-image/face-detail workflows are supported.

## Validation

The conversion was checked component-by-component against the pinned PyTorch
source before publication:

- UNet Core ML parity: 73.9 dB PSNR
- CLIP text encoder parity: 80.7 dB PSNR
- VAE decoder parity: 61.2 dB PSNR
- VAE encoder parity: 80.7 dB PSNR

An end-to-end 25-step generation was also completed with Apple's Swift Stable
Diffusion pipeline using CPU+GPU compute units.

## Recommended settings

- Scheduler: DPM-Solver++
- Steps: 25
- Guidance scale: 7.5
- Resolution: 512×512
- Batch size: 1 on iOS
- Device memory: 8 GB minimum in LocalMuse

## License and attribution

DreamShaper 8 is authored by Lykon and published under the
[CreativeML Open RAIL-M license](LICENSE). These files are modified from the
original by converting the model to compiled Core ML FP16 format and splitting
the UNet graph. No weights were retrained and no additional restrictions are
imposed. The original license and its use-based restrictions continue to apply.

This repository does not imply endorsement by Lykon, Stability AI, CompVis,
Runway, Apple or Hugging Face.