We have listed some frequently asked questions below. If you still have any further questions, please refer to the Contact page of the website to reach our organizers.
Q: How do I participate in the competition?
You just need to register by filling out the registration form in the Participate page. On the first page, submit your team name, contact email address, and each participant must read through and accept the Rules of the competition. On the second page, enter your each participant’s name, affiliation, and email address as formatted in the description of the form. After the registration is complete, you will receive an ID code that will help us identify your submissions or your team’s. You may then download the starter kit and get started!
Q: How many members can we have on a team?
There is no limit to the number of members on a team.
Q: I do not have a team. Can I still participate?
Yes, we welcome individuals as well as teams to participate.
Q: Can we use other datasets?
Using proprietary datasets or proprietary language models (ones only you and your affiliations have access to, or those behind a paywall) are NOT permitted.
However, open-sourced datasets are permitted for pre-training purposes only. For example, you can use open-sourced pre-trained transformers or language models on HuggingFace such as BERT.
You are not allowed to manually extend the given dataset for supervised training purposes since the focus of the competition is to be able to learn the task under low-resource setting.
Automatic data augmentation that creates variants of the provided dataset instances is permitted. You must however indicate so in the description of your submission and explain how it helps improve the performance of robustness of your model. For more information. read the “Use of Data” in the “Rules” section of this website.
Q: What do I need to make a submission?
Please carefully follow the instructions on Submissions.
Winning submissions will be required to include a brief description of your method.
Follow the submission protocol details on the website.
Q: What impact will my work have?
We created this competition to explore the use of NLP for improving the accessibility and usability of optimization solvers. We want to allow non-experts to describe their optimization problem in natural language and automatically translate the description to an input formulation that solvers can understand.
Q: Has anything like this been done before?
This dataset is unique and the tasks have never been done before.
Q: How was the data curated?
The data was created by 20 AI engineers and operation research experts over a 3-month period. Specific guidelines were followed to ensure the diversity of problem types and language patterns. Experts manually revised any incorrect or missing entries and an inter-annotator agreement was verified to ensure the quality of the annotation. Special care was taken to make sure that each problem did not include any inappropriate language or sensitive information like names of real people, products, or companies.
Q: What are the prizes?
We will provide a total monetary prize of $22000. The 5 best winning submission of each task will be awarded the following prizes (USD currency):
1st place: $6000
2nd place: $3000
3rd place: $1000
4th and 5th places: $500
Q: Can I add team members during the competition?
Adding team members during the competition is possible. To do so, you need to contact the organizers at this email address nl4opt@gmail