World Wires · story 45356 · corroborated · 1 source(s)

Cross-Interface Consistency Training Improves LLM Agent Robustness

This paper shows that cross-interface consistency training improves the robustness of LLM agents against semantics-preserving perturbations.

Open in the desk

Coverage

• Cross-Interface Consistency Training Improves LLM Agent Robustness arXiv: new preprints (security, AI, physics) · tier C · clear net · 2026-10-14 17:56 UTC

What this site indexes