File size: 3,581 Bytes
0fa0831 e079d73 0fa0831 e079d73 0fa0831 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 | ---
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
- q4
---
# Moebius ft_celebahq MLX q4
This folder contains a converted MLX version of the **ft_celebahq** Moebius checkpoint in **q4** form. `ft_celebahq` means the Moebius checkpoint fine-tuned on CelebA-HQ, for face and portrait inpainting.
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-q4` |
| Checkpoint meaning | Fine-tuned on CelebA-HQ for face and portrait inpainting. |
| 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 | `q4` |
| 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
4-bit MLX quantized export. The manifest selects a quantized UNet safetensors file. Standard q4 variants keep VAE encoder and decoder as regular f16 safetensors; special candidate variants can also select quantized VAE files and broader convolution packing.
- UNet precision mode: `q4`.
- MLX grouped quantization is used for supported linear layers; grouped quantized weights are loaded through the Moebius-MLX manifest/runtime.
- Quantization config: 4 bits, group size 64, standard MLX quantized mode.
- The VAE encoder and decoder remain regular f16 safetensors 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_quantized.safetensors` | 422.41 MB |
| VAE encoder | `vae_encoder.safetensors` | 68.34 MB |
| VAE decoder | `vae_decoder.safetensors` | 99.00 MB |
## Files in this folder
- `unet.safetensors`
- `unet_quantized.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.
|