Downloads · 30 days
62
33% of all-time downloads
krea-2-raw-mlx-q8
fxd0h88/krea-2-raw-mlx-q8
krea-2-raw-mlx-q8 is a text-to-image model from fxd0h88. Use it when you need an image from a text prompt. It is set up for mlx. The card lists the license as other.
An MLX q8 quantization of krea/Krea-2-Raw, the Krea 2 base checkpoint, for training LoRAs with mflux without the RAM to quantize the full-precision model locally. Krea recommends training on Raw; Turbo is the distille…
All-time downloads
187
Public
Parameters
12.8B
22.2 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors22.2 GB · 100%
Parameter types
How the weights are stored.
U3212.8B · 100%
Base models
Model card
From the Hugging Face model README
Krea 2 Raw (MLX, q8) for mflux LoRA training
An MLX q8 quantization of krea/Krea-2-Raw, the Krea 2 base checkpoint, for training LoRAs with mflux without the RAM to quantize the full-precision model locally. Krea recommends training on Raw; Turbo is the distilled inference checkpoint.
Quantized with mflux-save (q8). The text encoder is kept at bf16 (quantizing it degrades conditioning).
Usage
Point an mflux-train config at this repo ("model": "krea-2-raw", "model_path": "<this repo>"),
then use the trained LoRA on Krea 2 Turbo at inference.
- Task
- Text-to-Image
- Library
- mlx
- Type
- krea-2
- License
- other
- Created
- Jul 1, 2026
- Updated
- Jul 21, 2026