LLMs

This work proposes a retrieval-augmented LLM pipeline that identifies and optimizes problematic contract revisions using synthetic data, semantic retrieval, and acceptability-based alignment.