Evaluating and Improving LLM Self-Modeling

2026-08-31

AI Summary

研究通过创建基准和使用强化学习生成合成数据来提高大语言模型自我建模能力，解决其在回答自身行为问题上的局限。

Abstract

We study self-modeling: an LLM's ability to answer questions about its own behavior. We focus on verifiable behavioral questions, such as whether a prompt edit would change the model's final answer. To measure this capability, we introduce a benchmark that tests diverse types of self-modeling questions. Current models show non-trivial but limited self-modeling skill, and make systematic mistakes on simple counterfactual questions about their own behavior. To improve self-modeling skill, we develop a scalable synthetic-data pipeline that produces self-modeling training data, and show that reinforcement-learning can improve aggregate self-modeling skill across three open-source model families with some transfer to held-out tasks. These gains, however, do not seem to constitute introspection consistently: improved self-modeling may not arise from privileged access to the model's internal decision process.

Problem

• self-modeling
• verifiable behavioral questions
• systematic mistakes
• counterfactual questions

Innovation

• self-modeling
• benchmark
• reinforcement-learning
• synthetic-data