LEARNING CHESS AND NIM WITH TRANSFORMERS

Abstract

This project explores the application of transformer-based models in the domain of strategic board
games, specifically focusing on Chess and NIM. Leveraging the transformative power of advanced natural
language processing and sequence modeling offered by transformer architectures, the project aims to develop
intelligent agents capable of learning and mastering these complex games. By utilizing large-scale datasets and
fine-tuning transformer models, the system aims to understand and generate optimal moves in Chess and NIM,
demonstrating the adaptability of transformer technology beyond natural language tasks. The project
contributes to the intersection of artificial intelligence and board games, showcasing the potential of
transformer models in enhancing strategic decision-making processes and learning patterns in diverse gaming
environments. Through this exploration, the project sheds light on the versatility of transformers in non-textual
domains and their capacity to excel in strategic reasoning and problem-solving tasks.