Geometric Stability Analysis of LLMs in Chess: Beyond Accuracy

This research explores the stability of Large Language Models (LLMs) in a chess evaluation context, moving beyond simple accuracy metrics. The geometric stability analysis offers a deeper understanding of LLMs' decision-making processes.

• •Investigates the stability of LLMs for complex, strategic tasks.
• •Employs a geometric analysis approach to evaluate model behavior.
• •Suggests insights into improving LLM reliability in decision-making.