Datasets:
| license: mit | |
| language: | |
| - en | |
| pretty_name: W-Bench | |
| size: 10,000 instances (for evaluation) | |
| description: | | |
| The W-Bench is designed for the evaluation of image watermarking models, containing photographs sourced from the COCO, Flickr, and ShareGPT4V datasets. | |