---
base_model:
- stabilityai/stable-diffusion-xl-base-1.0
- madebyollin/sdxl-vae-fp16-fix
license: creativeml-openrail-m
---

# wmr model hub

This repository hosts the models [wmr](https://github.com/froggeric/gemini-watermark-and-synthid-remover)
downloads for its SynthID regeneration (`--synthid-attack regen`). wmr is a command-line tool that
removes visible watermarks from Gemini / Veo / NotebookLM generated media and scrubs the invisible
**SynthID** watermark via lossy SDXL img2img regeneration (the only SynthID attack the published
literature reports as validated, confirmed here against Google's official SynthID verifier).

wmr auto-downloads the files it needs on first use and caches them under `~/.cache/wmr/`; you do
not need to download anything from here manually. Pass `--regen-no-download` to refuse the network.
The SHA256 of every file is pinned in wmr's source and verified after download.

There are two regeneration backends, each with its own model set:

## 1. CoreML SDXL (macOS Apple Silicon, fast)

Used by `--synthid-attack regen` on macOS Apple Silicon (the default there). Native CoreML
`.mlpackage` directories converted from `apple/ml-stable-diffusion` tag 1.1.1, tarred for hosting.

| File | Size | SHA256 |
|------|------|--------|
| `coreml-sdxl-unet.mlpackage.tar.gz` | ~4.4 GB | `9101cadaefb5f98dad645ed81704ebe775ca173d926a61a4ed27a41d7e743f7a` |
| `coreml-sdxl-vae-encoder.mlpackage.tar.gz` | ~60 MB | `370232dd23330abe34c17b19d8b7c08f55c5938887ad1ef608b2bc0b4e000036` |
| `coreml-sdxl-vae-decoder.mlpackage.tar.gz` | ~88 MB | `f477d3ca98a19143d81c199c1d02d3ed1737d4638487042c71775016fe399424` |
| `empty_prompt_embeds.bin` | ~0.6 MB | `e27ab49bda70deda842a83afa00e67488533f881f5a359bf31b514c00a8038fe` |

The UNet + VAE are converted from Stability AI's SDXL base 1.0 + `madebyollin/sdxl-vae-fp16-fix`.
The empty-prompt embeddings are pre-baked CLIP outputs (the text encoders are not shipped).

## 2. CPU sdcpp SDXL (Linux, Windows, macOS Intel)

Used by `--synthid-attack regen` on Linux, Windows, and macOS Intel, and by `--regen-backend cpu`
everywhere. Run natively via [leejet/stable-diffusion.cpp](https://github.com/leejet/stable-diffusion.cpp)
(ggml), CPU-only in the release binaries.

| File | Size | SHA256 |
|------|------|--------|
| `sd_xl_base_1.0.safetensors` | ~6.5 GB | `31e35c80fc4829d14f90153f4c74cd59c90b779f6afe05a74cd6120b893f7e5b` |
| `sdxl_vae.safetensors` | ~335 MB | `235745af8d86bf4a4c1b5b4f529868b37019a10f7c0b2e79ad0abca3a22bc6e1` |

`sd_xl_base_1.0.safetensors` is the official Stability AI SDXL base 1.0 checkpoint (unchanged bytes;
its SHA is the HuggingFace LFS content oid, so it verifies identically to the upstream file).
`sdxl_vae.safetensors` is the fp16-stable VAE from `madebyollin/sdxl-vae-fp16-fix` (named
`sdxl_vae.safetensors` upstream). Both are mirrored here so the CPU path depends only on this repo.

## Licenses

- **SDXL base 1.0:** CreativeML Open RAIL++-M ([stabilityai/stable-diffusion-xl-base-1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0))
- **VAE fp16-fix:** MIT ([madebyollin/sdxl-vae-fp16-fix](https://huggingface.co/madebyollin/sdxl-vae-fp16-fix))
- **Conversion tool:** MIT ([apple/ml-stable-diffusion](https://github.com/apple/ml-stable-diffusion))