Is LLM-as-a-Judge Robust? Investigating Universal Adversarial Attacks on Zero-shot LLM Assessment.

Vyas Raina, Adian Liusie, Mark J. F. Gales

Browse the full EMNLP paper archive.