Update README.md
Browse files
README.md
CHANGED
@@ -4,3 +4,21 @@ widget:
|
|
4 |
- src: https://huggingface.co/carbon225/vit-base-patch16-224-hentai/resolve/main/samples/1.jpeg
|
5 |
- src: https://huggingface.co/carbon225/vit-base-patch16-224-hentai/resolve/main/samples/2.jpeg
|
6 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
- src: https://huggingface.co/carbon225/vit-base-patch16-224-hentai/resolve/main/samples/1.jpeg
|
5 |
- src: https://huggingface.co/carbon225/vit-base-patch16-224-hentai/resolve/main/samples/2.jpeg
|
6 |
---
|
7 |
+
|
8 |
+
# ViT for NSFW classification
|
9 |
+
|
10 |
+
## Model info
|
11 |
+
This is Google's [vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k)
|
12 |
+
finetuned for flagging images according to [vndb.org](https://vndb.org/d19) with 3 classes:
|
13 |
+
- safe
|
14 |
+
- suggestive
|
15 |
+
- explicit
|
16 |
+
|
17 |
+
## Training data
|
18 |
+
The model was trained on the vndb.org [database dump](https://vndb.org/d14)
|
19 |
+
using full size screenshots (`sf` in the database dump).
|
20 |
+
Because the dataset contains questionable images, I will not publish it.
|
21 |
+
|
22 |
+
## Intended use
|
23 |
+
The model can be used for flagging anime-style images for sexual content.
|
24 |
+
It can also be finetuned on other tasks related to anime images.
|