Mitigating Label Length Bias in Large Language Models

Large language models (LLMs) are increasingly utilized in finance and economics, where their ability to understand chronology is critical. A study tested this capability through various chronological ordering tasks, revealing that while models like GPT-4.1 and GPT-5 can maintain local order, they struggle with creating a consistent global timeline. The findings indicate a significant drop in exact match rates as task complexity increases, particularly in conditional sorting tasks, highlighting inherent limitations in LLMs' chronological reasoning.

Large language models (LLMs) are increasingly used for generating structured outputs, but existing methods often lack diversity in their results. A recent study confirms this limitation and proposes a new method to enhance output diversity through automaton-based structured generation. By utilizing automata traversal history, the method guides LLMs towards generating novel structural patterns. Evaluations indicate a significant improvement in both structural and content diversity while maintaining generation efficiency. A case study demonstrates its effectiveness in creating diverse test cases…