Archive

Three-axis uncertainty framework lifts code LLM reliability by 8.1 AUROC points

The work demonstrates that code-specific uncertainty estimation — rather than methods borrowed from natural language — meaningfully improves the ability to detect silently wrong programs, which is directly relevant to safe deployment of LLMs in agentic coding pipelines.