publications

2026

1. Are Latent Reasoning Models Easily Interpretable?

   Connor Dilgren and Sarah Wiegreffe

   In Conference on Language Modeling (COLM), 2026

   PDF Code
2. SecRepoBench: Benchmarking Code Agents for Secure Code Completion in Real-World Repositories

   Chihao Shen, Connor Dilgren, Purva Chiniya, and 3 more authors

   In LLM4Code Workshop at ICSE, 2026

   PDF Code Website