Viyog Weights
Finetuned classification checkpoints used in Viyog (separating adversarial from out-of-distribution inputs). 20 architectures (ResNet/DenseNet/ConvNeXt/Swin/ViT edge backbones) across CIFAR-100, CIFAR-10, GTSRB.
Source du modèle
- Type
- Other
- Taille du fichier
- 13,6–655 MB
- Version
- vConvnextv2 base
- Mis à jour
- 22 oct. 2026
- Source
- Hugging Face
Description de la source
Viyog — finetuned backbones
Finetuned classification checkpoints used in Viyog (separating adversarial from out-of-distribution inputs). 20 architectures (ResNet/DenseNet/ConvNeXt/Swin/ViT
- edge backbones) across CIFAR-100, CIFAR-10, GTSRB.
Layout — state_dict .pth files:
- root: the 4 original CIFAR-100 backbones (convnextv2_base, swin_tiny, vit_b, tf_efficientnetv2_l)
cifar100/,cifar10/,gtsrb/: per-dataset checkpoints for every architecture
Package: pip install viyog · code: https://github.com/amanyagami/viyog ·
demo: https://huggingface.co/spaces/amanyagami/viyog
Sources
1 source
Sources
1 sourceVérifié 22 oct.
Download files
Artefacts du modèle
34 artefacts
Artefacts du modèle
34 artefactsswin_tiny_patch4_window7_224_cifar100.pth
pth · 172 MB · SHA-256 a843101bed5c…0e81 · Hugging Face
Download filetf_efficientnetv2_l_cifar100.pth
pth · 454 MB · SHA-256 0cf62a4041db…9f1d · Hugging Face
Download fileExtraits de sources
2 extraits
Extraits de sources
2 extraits--- license: mit library_name: timm tags: - image-classification - adversarial-robustness - ood-detection - viyog --- # Viyog — finetuned backbones Finetuned classification checkpoints used in **Viyog** (separating adversarial from out-of-distribution inputs). 20 architectures (ResNet/DenseNet/ConvNeXt/Swin/ViT + edge backbones) across **CIFAR-100, CIFAR-10, GTSRB**. **Layout** — `state_dict` `.pth` files: - root: the 4 original CIFAR-100 backbones (convnextv2_base, swin_tiny, vit_b, tf_efficientnetv2_l) - `cifar100/`, `cifar10/`, `gtsrb/`: per-dataset checkpoints for every architecture Package: `pip install viyog` · code: https://github.com/amanyagami/viyog · demo: https://huggingface.co/spaces/amanyagami/viyog
Source context: 0 downloads · 0 likes · Pipeline image-classification · Library timm · Repo amanyagami/viyog-weights
