Limits of Emergent Reasoning of Large Language Models in Agentic Frameworks for Deterministic Games
This research investigates the performance collapse of Large Language Models in agentic frameworks when exceeding specific perplexity thresholds, emphasizing...