Translation. Region: Russian Federation –
Source: State University Higher School of Economics – State University Higher School of Economics –
From October 13 to 20, the HSE University held a hackathon “HSE AY Assistant Hakk: Pothon”, organized Faculty of Computer Science And Center for Artificial Intelligence HSE. 89 student teams from the country’s leading universities competed for prizes.
The challenge asked participants to use LLM and other machine learning algorithms to create an AI assistant that would help student programmers and developers solve Python problems by understanding where an error was made in their code and offering step-by-step explanations and recommendations for fixing it without explicitly providing the corrected code.
The competition lasted seven days, five of which were held online on the DS Works platform from the cloud technology provider Cloud.ru. The opening and final days were held in the HSE building on Pokrovsky Boulevard.
16 teams that were among the top ranking teams based on the quantitative metrics obtained for their solution were allowed to defend their projects.
The jury included experts from the Center for Artificial Intelligence,Sber’s basic department “Financial technologies and data analysis”, teachers of the Faculty of Computer Science and the projectDate Culture.
The victory was won by the team “MISIS Computer” from the University of Science and Technology MISIS. The second place was taken by the team MMG from the Financial University under the Government of the Russian Federation. The bronze was won by the team Selling Pandas, consisting of first-year students of the program“Applied Mathematics and Computer Science” HSE Faculty of Computer Science. The winners received cash prizes of 250,000, 200,000 and 150,000 rubles, respectively.
Vice-Rector of the National Research University Higher School of Economics Sergey Roshchin and Head of the Center for Artificial Intelligence Alexey Masyutin awarded the winners and shared their opinions on the hackathon.
Sergey Roshchin
— The Higher School of Economics is developing in order to respond to the challenges that occur in technology, business processes and various spheres of human life. The big digital world is a new reality in which we must learn to live and interact with artificial intelligence technologies. That is why we are actively building education so that absolutely all of our students, regardless of their field of study, master digital competencies. The Higher School of Economics is a recognized leader in training personnel in the field of AI. But we are going further and setting ourselves more ambitious goals, including automating routine tasks related to the development of AI technologies. We chose the hackathon format because it allows students to test and prove themselves by solving important and new problems. The competition participants spent a whole week developing a solution, and during this time, I am sure, they managed not only to find a worthy solution, but also to gain experience and develop their human capital. And the results obtained are a contribution to our digital future.
Alexey Masyutin
— “HSE AI Assistant Hack: Python” is unique in that it combines several important areas of HSE:teaching digital skills to students of all disciplines Andcreation of AI assistants, including in the educational process.
This time we managed to hold a real hackathon, when there is no pre-defined pipeline on how to get a solution.
Participants had the opportunity to adapt language models to help the student step-by-step understand the problem while writing code, rather than simply providing a ready-made solution.
Based on the results of the defenses, variants were proposed with enrichment with synthetic data, and with advanced aroma engineering, and with RAG approaches, and even with additional LLM training. This exceeded our expectations.
We intend to use the best solutions for implementation in HSE Smart LMS and provide an AI assistant for both students and teachers when learning Python.
The teams that won prizes also shared their impressions.
First place – the team “MISIS Computer”
— The hackathon was really cool, our team has a lot of experience participating in hackathons, and we know firsthand what a quality organization is. This hackathon had a really great organization, a good leaderboard, clear metrics and, most importantly, active organizers and experts who promptly answered questions and were in touch. Special thanks to Maxim — for his openness and cool expertise on the task. The results of the hackathon will not pass by HSE, the organizers are going to implement the best solutions in the university systems. Thanks to the organizers, experts and participants for such an interesting competition! We will be back!
Second place – MMG team
— Our team took part in the hackathon with great pleasure. We were especially pleased that it was technical and research-based, without a business component, and that the task was very relevant and in demand. Despite the fact that for most of us such a deep dive into working with LLM was the first time, we successfully coped with it, mastered new technologies and gained valuable experience.
It was an honor for us to represent the Financial University at this event. We would like to express special gratitude to the organizers for their constant support. They quickly resolved any difficulties that arose and were always in touch, which created comfortable conditions for work. We are glad that we were able to take part in this hackathon and take such a high place!
Third place – Selling Pandas team
— This was our first hackathon, which essentially consisted entirely of using fine-tuning LLMs (large language models). It was not easy, but it is precisely such competitions that bring the most pleasure, when you need to create a solution in a limited time in a practically new area for you.
By the end of the hackathon, we didn’t even want it to end, as there were still ideas that we hadn’t managed to implement. We were also very pleased with the organization of the hackathon: they always provided clear information and promptly answered all questions. We are very happy that we managed to take the prize place.
Please note: This information is raw content directly from the source of the information. It is exactly what the source states and does not reflect the position of MIL-OSI or its clients.