ModelWars

ModelWars is a multi-stage Machine Learning competition where models fight, strategies clash, and adaptability is the key to survival. Battle through dataset challenges, token-based decisions, adversarial sabotage, and model explanation rounds to prove your machine learning skills. Teams build a machine learning model that can withstand corrupted data, adapt to missing essential features, and clearly justify the reasoning behind their modelling choices. Where Models Fight, Minds Win! Guidelines: Eligibility: Open to students from all colleges and universities. Team Size: 1 - 3 participants per team. Inter-College Teams: Allowed. Pariticipants from different colleges may form a team. Inter-Specialization Teams: Allowed. Participants from differents branches/specializations may form a team. Competition Format: The competition consists of three rounds: The Dataset Games: Model development, progressive dataset unlocking, token management, sidequests, and leaderboard competition. Model Kombat: Adversarial dataset sabotage and model adaptation. Explainfinity War: In-person explanation and defence of modelling approaches and decisions. Dataset Access: Teams begin with 30% of the dataset and can unlock additional portions progressively: 30% → 45% → 70% → 100%, subject to the required model performance threshold. Tokens: Teams use tokens for private score checks and leaderboard submissions. Additional tokens can be earned through sidequests and bonuses. Qualification: Only the top-ranked teams from Round 1 qualify for Round 2. The highest-ranked teams from Round 2 advance to the final round. Round 3 Presentation: Finalists must explain their modelling approach, justify their decisions and trade-offs, and discuss how they adapted to the sabotage introduced in Round 2. Duration: Round 1 - 75 minutes; Round 2 - 75 minutes; Round 3 - 45 minutes. Date and Time: 25th September 2026, 10:00 AM - 2:00 PM. Venue: SNU Chennai, AB3 Lab 14 and Lab 15, Second Floor.