README.md

---
license: creativeml-openrail-m
base_model: RunDiffusion/Juggernaut-X-Hyper
pipeline_tag: text-to-image
tags:
  - core-ml
  - coreml
  - stable-diffusion
  - apple-silicon
  - ios
  - quantized
---

# Juggernaut X Hyper — Core ML (8-bit)

![demo](demo.png)

<sub>**Generated on-device from this exact Core ML build** (6 steps, guidance 2.0, trailing timestep spacing, 1024x1024, seed 42 — 12 s on an M3 Ultra).</sub>

**Photorealism + speed** — Hyper-SD low-step variant of the Juggernaut X line.

Core ML conversion for Apple silicon (iOS / iPadOS / macOS, Neural Engine), built with Apple's
[ml-stable-diffusion](https://github.com/apple/ml-stable-diffusion) for
[mindfire-image](https://github.com/Gatcha-man/mindfire-image).

## Original model

Converted from **[RunDiffusion/Juggernaut-X-Hyper](https://huggingface.co/RunDiffusion/Juggernaut-X-Hyper)** — go there for the
original weights, full model card and licence.

## Demo prompt

The prompt and settings used for this model's demo image (also the reference example shipped
in mindfire-image):

**Prompt**
```
Cinematic mid shot photo of an astronaut walking through a neon-lit Tokyo alley at night, hyperdetailed photography, skin details, shallow depth of field
```

| Setting | Value |
|---|---|
| Steps | 6 |
| Guidance (CFG) | 2.0 |
| Size | 1024x1024 |

Hyper-SD: few steps at low guidance. The Unet is **chunked** (`UnetChunk1/2.mlmodelc`) to fit Neural Engine per-model limits.

## Modifications from the base model

Converted from PyTorch/diffusers to Core ML (`.mlmodelc`) and **quantized to 8-bit palettized**
weights. No fine-tuning — behaviour tracks the base model, though quantization can shift outputs
slightly.

## Licence

Inherited from the base model: `creativeml-openrail-m`. This carries the OpenRAIL **use-based restrictions
(Attachment A)**, which bind you as a downstream user of this conversion exactly as they do for the
base model. Read the base model's licence before use or redistribution.