Ming-Image-0.1-Design INT4 ConvRot
Repackaged model files for ComfyUI and others, with INT4 ConvRot converted for testing in Radiant Canvas for fast macOS inference: Radiant Canvas on the App Store.
The original model is inclusionAI/Ming-Image-0.1-Design.
File
ming_image_0.1_design_int4_convrot.safetensors: the diffusion transformer in ComfyUI's native ConvRot W4A4 format.
For ComfyUI, place the file in:
ComfyUI/models/diffusion_models/
Use a recent ComfyUI build with Ming-Image and native INT4 ConvRot support. The text model, connector, MLP, and VAE are not included in this repository. Use the original model files or another compatible ComfyUI package for those components.
Conversion details
- Source revision:
208087ada1486931692c1896f38d4cd16ff3df82 - Source format: BF16, five transformer shards, 519 tensors
- Quantized layers: 102 feed-forward weights
- ConvRot group size: 256
- Quantization group size: 64
- Sensitive attention, input, output, modulation, and normalization layers are kept in BF16 for compatibility.
- Output size: 6,296,983,672 bytes
- SHA-256:
faaa099bfd8ecc704bbf2655a9bf78d80bf57f26df172bedd6f07d27cb114d36
This is a test conversion. Check output quality for your own prompts and workflows before relying on it.
License
The original model is released under the MIT License. A copy is included in this repository.