| --- |
| license: apache-2.0 |
| library_name: mlx |
| pipeline_tag: image-to-image |
| base_model: hustvl/Moebius |
| tags: |
| - mlx |
| - image-inpainting |
| - inpainting |
| - diffusion |
| - moebius |
| - ft_celebahq |
| - f16 |
| --- |
| |
| # Moebius ft_celebahq MLX f16 |
| |
| This folder contains a converted MLX version of the **ft_celebahq** Moebius checkpoint in **f16** form. |
| |
| Original upstream model: [hustvl/Moebius](https://huggingface.co/hustvl/Moebius) |
| Original source repository: [hustvl/Moebius](https://github.com/hustvl/Moebius) |
| |
| Moebius checkpoint fine-tuned on CelebA-HQ. This variant is intended for face and portrait inpainting distributions. |
| |
| ## Identity |
| |
| | Field | Value | |
| |---|---| |
| | Variant name | `ft_celebahq-f16` | |
| | Original Moebius checkpoint family | `ft_celebahq` | |
| | Original checkpoint type | CelebA-HQ fine-tune | |
| | Source PyTorch checkpoint | `Moebius-Models/ft_celebahq/diffusion_pytorch_model.bin` | |
| | MLX precision / quantization label | `f16` | |
| | Image size | 512 x 512 | |
| | Latent size | 64 x 64 | |
| | Latent channels | 4 | |
| | Mask channels | 1 | |
| | Conditioning IDs | 20 | |
| | VAE scaling factor | 0.13025 | |
| | Noise offset | 0.0357 | |
|
|
| ## Quantization |
|
|
| Float16 MLX export. UNet, VAE encoder, and VAE decoder are stored as regular `*.safetensors` arrays converted to float16. This roughly halves the f32 storage while keeping the same non-quantized module layout. |
|
|
| - No MLX quantized module packing is used for this variant. |
|
|
| ## Manifest-selected deployment files |
|
|
| These are the files selected by `manifest.json` when the Moebius-MLX runtime loads this variant. |
|
|
| | Component | File | Size | |
| |---|---|---:| |
| | UNet | `unet.safetensors` | 452.55 MB | |
| | VAE encoder | `vae_encoder.safetensors` | 68.34 MB | |
| | VAE decoder | `vae_decoder.safetensors` | 99.00 MB | |
|
|
| ## Files in this folder |
|
|
| - `unet.safetensors` (selected by manifest) |
| - `vae_decoder.safetensors` (selected by manifest) |
| - `vae_encoder.safetensors` (selected by manifest) |
| - `manifest.json` (runtime metadata and file selection) |
|
|
| A minimal runtime package needs `manifest.json` and the manifest-selected files above. Extra source or fallback files are optional and are not required for inference. |
|
|
| ## Runtime expectations |
|
|
| This is not a Transformers or Diffusers-native checkpoint. It is intended for the Swift/MLX runtime in [Moebius-MLX](https://github.com/kylehowells/Moebius-MLX). The runtime reads `manifest.json`, loads the selected safetensors files, builds the Moebius UNet and VAE modules, and runs the DDIM inpainting pipeline. |
|
|
| Pipeline constants must match the manifest: |
|
|
| - DDIM scheduler: `scaled_linear`, beta start 0.00085, beta end 0.012, 1000 train timesteps, clip sample false |
| - 512 x 512 image resolution and 64 x 64 latent resolution |
| - 9-channel UNet input: noisy latent, mask, and masked-image latent |
| - VAE scaling factor 0.13025 |
|
|
|
|
| ## Attribution |
|
|
| Moebius was released by the original authors as [hustvl/Moebius](https://huggingface.co/hustvl/Moebius). This folder is a format conversion and/or quantized MLX packaging of the original PyTorch weights, not a newly trained model. |
|
|
