Instructions to use ashen-sensored/wd-eva02-tagger-2026-canary with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- timm
How to use ashen-sensored/wd-eva02-tagger-2026-canary with timm:
import timm model = timm.create_model("hf_hub:ashen-sensored/wd-eva02-tagger-2026-canary", pretrained=True) - Notebooks
- Google Colab
- Kaggle
WD EVA02 Tagger 2026 Canary
Training Data cutoff: 20260518, danbooru ID 11403645 Tag metadata source: u-haru/danbooru-tags-20260518.
Reference: base model SmilingWolf/wd-eva02-large-tagger-v3
Model Bio
- 5,999 new tags added (2,205 character + 3,794 general), 16,473 total.
- Trained on images with IDs above 7,220,105(Original model knowledge cutoff), modulo 0000-0899
- Validated modulo 0900-0949
- Per-label operating-point recentering post-training
Validation results
P=R: threshold = 0.6094, F1 = 0.5416
Inference code examples
For timm: https://github.com/neggles/wdv3-timm
import timm
from safetensors.torch import load_file
model = timm.create_model("eva02_large_patch14_448", pretrained=False, num_classes=16473)
model.load_state_dict(load_file("model.safetensors"), strict=True)
model.eval()
- Downloads last month
- 76
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support