ViT for NSFW classification
Model info
This is Google's vit-base-patch16-224-in21k finetuned for flagging images according to vndb.org with 3 classes:
- safe
- suggestive
- explicit
Training data
The model was trained on the vndb.org database dump
using full size screenshots (sf in the database dump).
The dataset can be loaded from carbon225/vndb_img.
Intended use
The model can be used for flagging anime-style images for sexual content. It can also be finetuned on other tasks related to anime images.
- Downloads last month
- 101
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.