How to Participate?
Go to Dynabench.org and either log in to your account or create a new one.
Click on the Adversarial Nibbler challenge.
Start experimenting with safe looking prompts that you think will cause the model to generate unsafe images
Iterate on step 3 until you've identified an unsafe image you would like to submit
Provide the requested information about your prompt and image
Repeat steps 3-5 in order to submit multiple prompt-image pairs.
You can perform these steps within a single session or across multiple submission sessions during the duration of the challenge.
Please be aware that you are limited to 50 sets of image generations per day. If you reach this limit, come back the following day.
Working with adversarial data can be challenging. The prompts that you create and the images that are generated may be upsetting. We've put together a list of resources that are available to you. Please don't hesitate to reach out via email (firstname.lastname@example.org) or the slack group (adversarial-nibbler.slack.com) if you prefer to speak with one of the organizers directly.
Handling Traumatic Imagery: Developing a Standard Operating Procedure: Practical tips for ensuring your own well-being. We encourage you to consider employing any of the strategies detailed on the site, including taking breaks and talking to others working on the same (or a similar) task.
The Vicarious Trauma Toolkit: A list of over 500 resources spanning podcasts, videos, research articles, and help websites.
Contact the organizers at email@example.com or join our slack channel at adversarial-nibbler.slack.com