Q1: Are there any restrictions on the team formation?
A1: For each team, there is no preset limitation on the number of team members, their roles, or their affiliations.
Q2: Can we add/remove members and/or advisors?
A2: No. You are not allowed to add/remove members and/or advisors after successful registration.
Q3: How should we submit the implementations, via Email?
A3: We set up a Google form. All submissions should be done via the provided Google Form.
Q4: Can we use the ideas in the published papers?
A4: Yes. Typically, the contestants do not need to submit any document to describe their algorithms. The contestants only need to submit their source code.
Q5: Do we need to fine-tune models by ourselves?
A5: No. Although we would appreciate it if you submit the model fine-tuned with your dataset, for both phases, you can submit the data samples you collected/generated and the labels you created without the corresponding fine-tuned model. We will fine-tune the model using our provided codebase during evaluation.
Q6: Is the contest only open to academia?
A6: No. Participants from both industry and academia are welcome. However, please be aware that we do anticipate that all implementations should be open-source after the contest.
Q7: In Phase I, are there any limitations on how to label the submitted data samples?
A7: No, in Phase I, there are no limitations on how labels of the submitted data samples are generated.
Q8: In Phase II, can I keep the default labels from the provided comprehensive dataset?
A8: Yes, you can keep the default labels.
Q9: Can I design my own fine-tuning technique? How will you fine-tune the model?
A9: We will use the commonly used fine-tuning settings to fine-tune the model, detailed settings can be found in the starting toolkit. To ensure a fair comparison, we do not allow customized fine-tuning setting at this time.