General Rules

• We welcome participation from everyone who satisfies the eligibility conditions outlined in the Eligibility Section.

• A team can be either an individual or a group of individuals working together during the entire duration of the competition.

• Each team may have up to 3 folders in their submission cloud directory at a time. They will be removed upon evaluation, and more submissions may be uploaded again.

• To receive the awards from this competition, winners must attend the NeurIPS workshop.

• Organizers have a right to update the rules in the case of unforeseen circumstances (i.e. a winner is unable or unwilling to accept the prize)

• By your submission, you are allowing us, the organizers to:

  • Access the code and trained model;

  • Test and analyze the results and review the source code to ensure compliance to the rules of the competition;

  • Promote the results and findings;

  • Publish information about the winning methods and solutions in a technical report after the competition;

  • Leverage the methods with no additional compensation other than the prizes of this competitions.

Eligibility

• Individuals must be eligible to participate in this competition.

• To be deemed eligible, the following requirements must be met:

  • You are not a member of the organizing committee nor a family member of anyone on the organization committee;

  • You were not involved in the planning of the competition;

  • You are not a direct collaborator with anyone on the organizing committee;

  • You agree to abide by the rules detailed in this document;

  • If you are on the winning team, to receive the prize (except with parent/guardian written consent), you are at least 18 and the age of majority in your place of residence.

Use of Data

• Individuals must understand and abide by the MIT license.

• Individuals must understand and abide by the anonymity agreement:

  • Individuals are required to share a valid email address with the organizers but may keep their other personal information private. However, to receive the prize, winning individuals must disclose their name, professional affiliation, and address;

  • If participants want to submit their methods for publication and need to maintain anonymity of their work, we are not responsible for maintaining the anonymity;

  • For the benefit of the community, winning teams will be required to describe their methods in detail in a written document.

• Participants may not use proprietary datasets or proprietary models. However, you may use open-sourced models such as pre-trained transformers or language models.

• Participants may not expand the dataset manually since the focus of this competition is to work with a smaller dataset. However, intelligent data augmentation or open-sourced datasets to pre-train the transformers or language models are permitted.

• The organizers will inspect the data reader/loader provided by the participants.

• You may not use the dev set for training as part of the goal of this competition is to evaluate how well each approach generalizes to an unseen domain.

  • The winning submissions will be inspected and tested for reproducibility to ensure that the dev set was not utilized for training purposes