When Do LLM Agents Treat Surface Noise Differently from Semantic Noise? A 68-Cell Measurement Study with a Held-Out Trace-Level Validation
Published in Submitted to EMNLP 2026, 2026
A controlled 68-cell measurement study comparing LLM-agent responses to surface and semantic noise, complemented by held-out trace-level validation.
Recommended citation: L. Zhang and J. Guo, "When Do LLM Agents Treat Surface Noise Differently from Semantic Noise? A 68-Cell Measurement Study with a Held-Out Trace-Level Validation," submitted to EMNLP 2026.