defect-aware-MAE is an open-weight model from Shrey Patel. Its published files total 350.6 MB.
A Vision Transformer encoder pretrained without labels on steel-surface images with a masked autoencoder adapted to texture defects. Frozen features are clustered into defect categories.
Model Card
A Vision Transformer encoder pretrained without labels on steel-surface images with a masked autoencoder adapted to texture defects. Frozen features are clustered into defect categories. No class labels, bounding boxes or detection objective are used in pretraining, and no contrastive loss. Encoder: 224×224 grayscale input, 16×16 patches (196 tokens), 6 pre-norm Transformer layers, 12 heads, 768-d. The decoder in the MAE checkpoint has 2 layers and a 36-d output head: Decoder(depth=2) with dec.pred = torch.nn.Linear(256, 36). NEU-DET (hot-rolled steel strip, 6 defect classes, 1,800 images). Pretraining used only the 1,440-image training split. The 360-image validation split was held out…
Excerpt from the card by Shrey Patel.
Identity and Version
- Repository
- Godseye1311/defect-aware-MAE
- Publisher
- Shrey Patel
- Task
- Not stated by the source
- Modality
- Other
- Library
- pytorch
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- 2edccd13b24cb969543dba963622739d355e87c5
- First published
- 2026-10-14
- Last updated
- 2026-10-14
Files and Weights
8 files, 350.6 MB in total. The weights are 2 files totalling 350.4 MB in pth.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| defect_aware_mae_seed0.pth | Weights | 178.9 MB | e91250099106 |
| simclr_baseline_seed0.pth | Weights | 171.5 MB | 483b0e170839 |
| defect_aware_mae_seed0_args.json | Configuration | 273 B | — |
| defect_aware_mae_seed0_history.json | Configuration | 134.2 KB | — |
| model.py | Configuration | 8.8 KB | — |
| simclr_baseline_seed0_args.json | Configuration | 264 B | — |
| README.md | Documentation | 4.8 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 350.4 MB
Released by Shrey Patel through its official repository on Hugging Face.
Built From
- Trained on (disclosed) NEU-DET
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 350.4 MB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.