Downloads · 30 days
0
FUSED
aonuzhdin/FUSED
FUSED is a machine learning model from aonuzhdin. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for pytorch. The card lists the license as mit.
Weights for FUSED: forensic SparseViT + frozen CLIP-ConvNeXt-XXL fused by a sparse MoE, with a joint detection and localization head.
Access
Public
Updated Aug 29, 2026
Repo size
841 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth841 MB · 100%
At a glance
- Library
- pytorch
- License
- mit
- Access
- Public
- Created
- Aug 29, 2026
- Updated
- Aug 29, 2026
- SHA
- b7dfe0cb
Model card
From the Hugging Face model README
FUSED checkpoints
Weights for FUSED: forensic SparseViT + frozen CLIP-ConvNeXt-XXL fused by a sparse MoE, with a joint detection and localization head.
ConvNeXt-XXL is not stored here. load_fused(..., strict=True) fills it from timm (convnext_xxlarge.clip_laion2b_soup).
Files
| File | Size | Use |
|---|---|---|
fused_opensdi.pth | ~306 MB | FUSED trained on OpenSDID SD1.5. Pass as checkpoint= for paper eval / inference. |
fused_sofake.pth | ~306 MB | FUSED trained on sqrt-balanced So-Fake. Same load path. |
sparsevit_pretrained.pth | ~193 MB | SparseViT forensic initialization. Pass as trainer.checkpoint= when training. |
Load
from huggingface_hub import hf_hub_download
from fused.checkpoint import load_fused
from fused.model import build_model
path = hf_hub_download("aonuzhdin/FUSED", "fused_opensdi.pth")
model, ckpt = load_fused(path, map_location="cpu")
Or from the repo:
uv run python scripts/download_weight.py
uv run python inference.py checkpoint=weights/fused_opensdi.pth input=image.jpg
- Library
- pytorch
- License
- mit
- Created
- Aug 29, 2026
- Updated
- Aug 29, 2026