FAQ

Frequently Asked Questions

Model Requirements and Recommendations

Q: What are the computational requirements for models? A: Models should be able to run on a single GPU with 20 GB of memory. For testing purposes, use a batch size of one subject.

Team Composition

Q: Can I be part of multiple teams? A: No, participants can only be part of one team. Multiple accounts for the same person are not allowed.

Q: Is there a limit on team size? A: No, there is no size limit for teams.

Q: Who should be designated as the team leader? A: Each team should identify one leader who will be responsible for presenting the solution if the team wins and for coordinating prize sharing among team members.

Submissions and Model Usage

Q: Can I use external datasets in addition to the provided data? A: Yes, you can use other datasets, but you must clearly mention and document any additional data sources used in your submission.

Q: Are pre-trained foundation models allowed? A: Yes, you are more than welcome to use existing foundation models, but you must clearly mention and document which pre-trained models were used and how they were fine-tuned in your submission.

Data Access

Q: Where can I access the competition dataset? A: The HBN-EEG dataset is available on NEMAR.org and Amazon S3. For detailed download links and access information, please visit dataset page.

Q: Are there smaller datasets available for initial exploration? A: Yes, mini datasets for each release are available for download. You can access these either as zip files or via Google Drive links provided on the dataset page.

For additional questions not covered here, please visit the challenge Google Group to see if other participants have faced similar issues or to ask questions (group is public, but requires membership to post).