engine renamed: KilnImage → Mirage
Browse files
README.md
CHANGED
|
@@ -21,14 +21,14 @@ base_model: baidu/ERNIE-Image-Turbo
|
|
| 21 |
# ERNIE-Image-Turbo — iOS bundle
|
| 22 |
|
| 23 |
<p align="center">
|
| 24 |
-
<a href="https://github.com/haplollc/
|
| 25 |
<a href="https://huggingface.co/baidu/ERNIE-Image"><img alt="Upstream" src="https://img.shields.io/badge/Upstream-baidu%2FERNIE--Image-blue" /></a>
|
| 26 |
<img alt="License" src="https://img.shields.io/badge/license-Apache--2.0-lightgrey" />
|
| 27 |
<img alt="Params" src="https://img.shields.io/badge/params-8B-purple" />
|
| 28 |
<img alt="Distill" src="https://img.shields.io/badge/distill-Turbo-green" />
|
| 29 |
</p>
|
| 30 |
|
| 31 |
-
A mobile-friendly bundle of **ERNIE-Image-Turbo** — Baidu's 8B single-stream DiT, distilled for fast inference, with state-of-the-art text rendering quality among open-weight models. Bundled with Ministral 3B text encoder + Flux VAE for on-device inference via [**
|
| 32 |
|
| 33 |
ERNIE-Image-Turbo is particularly strong at:
|
| 34 |
- **Photorealism** at 1024×1024
|
|
@@ -45,10 +45,10 @@ ERNIE-Image-Turbo is particularly strong at:
|
|
| 45 |
|
| 46 |
Total bundle: **~5.9 GB**. Total GPU residency: ~7 GB. **iPhone 16 Pro / 17 Pro / Mac** territory.
|
| 47 |
|
| 48 |
-
## Quick start (
|
| 49 |
|
| 50 |
```swift
|
| 51 |
-
import
|
| 52 |
|
| 53 |
let docs = FileManager.default.urls(for: .documentDirectory, in: .userDomainMask)[0]
|
| 54 |
|
|
@@ -79,7 +79,7 @@ If you need **text inside images** that actually renders correctly (signs, label
|
|
| 79 |
| Text encoder | [unsloth/Ministral-3-3B-Instruct-2512-GGUF](https://huggingface.co/unsloth/Ministral-3-3B-Instruct-2512-GGUF) | Apache 2.0 |
|
| 80 |
| VAE | [ffxvs/vae-flux](https://huggingface.co/ffxvs/vae-flux) (re-host of FLUX.1's `ae.safetensors`) | FLUX-1-dev-non-commercial |
|
| 81 |
|
| 82 |
-
## Performance (
|
| 83 |
|
| 84 |
| Device | 1024² @ 8 steps |
|
| 85 |
|---|---|
|
|
@@ -89,4 +89,4 @@ If you need **text inside images** that actually renders correctly (signs, label
|
|
| 89 |
|
| 90 |
## Built by
|
| 91 |
|
| 92 |
-
[Haplo](https://haplo.app) · [
|
|
|
|
| 21 |
# ERNIE-Image-Turbo — iOS bundle
|
| 22 |
|
| 23 |
<p align="center">
|
| 24 |
+
<a href="https://github.com/haplollc/Mirage"><img alt="Mirage" src="https://img.shields.io/badge/Runs%20on-Mirage-orange" /></a>
|
| 25 |
<a href="https://huggingface.co/baidu/ERNIE-Image"><img alt="Upstream" src="https://img.shields.io/badge/Upstream-baidu%2FERNIE--Image-blue" /></a>
|
| 26 |
<img alt="License" src="https://img.shields.io/badge/license-Apache--2.0-lightgrey" />
|
| 27 |
<img alt="Params" src="https://img.shields.io/badge/params-8B-purple" />
|
| 28 |
<img alt="Distill" src="https://img.shields.io/badge/distill-Turbo-green" />
|
| 29 |
</p>
|
| 30 |
|
| 31 |
+
A mobile-friendly bundle of **ERNIE-Image-Turbo** — Baidu's 8B single-stream DiT, distilled for fast inference, with state-of-the-art text rendering quality among open-weight models. Bundled with Ministral 3B text encoder + Flux VAE for on-device inference via [**Mirage**](https://github.com/haplollc/Mirage).
|
| 32 |
|
| 33 |
ERNIE-Image-Turbo is particularly strong at:
|
| 34 |
- **Photorealism** at 1024×1024
|
|
|
|
| 45 |
|
| 46 |
Total bundle: **~5.9 GB**. Total GPU residency: ~7 GB. **iPhone 16 Pro / 17 Pro / Mac** territory.
|
| 47 |
|
| 48 |
+
## Quick start (Mirage)
|
| 49 |
|
| 50 |
```swift
|
| 51 |
+
import Mirage
|
| 52 |
|
| 53 |
let docs = FileManager.default.urls(for: .documentDirectory, in: .userDomainMask)[0]
|
| 54 |
|
|
|
|
| 79 |
| Text encoder | [unsloth/Ministral-3-3B-Instruct-2512-GGUF](https://huggingface.co/unsloth/Ministral-3-3B-Instruct-2512-GGUF) | Apache 2.0 |
|
| 80 |
| VAE | [ffxvs/vae-flux](https://huggingface.co/ffxvs/vae-flux) (re-host of FLUX.1's `ae.safetensors`) | FLUX-1-dev-non-commercial |
|
| 81 |
|
| 82 |
+
## Performance (Mirage, rough)
|
| 83 |
|
| 84 |
| Device | 1024² @ 8 steps |
|
| 85 |
|---|---|
|
|
|
|
| 89 |
|
| 90 |
## Built by
|
| 91 |
|
| 92 |
+
[Haplo](https://haplo.app) · [Mirage on GitHub](https://github.com/haplollc/Mirage)
|