Towards a Probabilistic Framework for Analyzing and Improving LLM-Enabled Software
This work addresses the problem of developing robust LLM-enabled systems for software engineers, offering a principled but incremental approach to system analysis and improvement.
The paper tackles the challenge of ensuring reliability and verifiability in LLM-enabled software systems by proposing a probabilistic framework that models distributions over semantically equivalent outputs to analyze and improve Transference Models. In a case study on autoformalization, this approach identified weaknesses and guided improvements, leading to more reliable and interpretable outputs.
Ensuring the reliability and verifiability of large language model (LLM)-enabled systems remains a significant challenge in software engineering. We propose a probabilistic framework for systematically analyzing and improving these systems by modeling and refining distributions over clusters of semantically equivalent outputs. This framework facilitates the evaluation and iterative improvement of Transference Models--key software components that utilize LLMs to transform inputs into outputs for downstream tasks. To illustrate its utility, we apply the framework to the autoformalization problem, where natural language documentation is transformed into formal program specifications. Our case illustrates how distribution-aware analysis enables the identification of weaknesses and guides focused alignment improvements, resulting in more reliable and interpretable outputs. This principled approach offers a foundation for addressing critical challenges in the development of robust LLM-enabled systems.