paper-with-me

홈 › Papers

AdvJudge-Zero: Binary Decision Flips in LLM-as-a-Judge via Adversarial Control Tokens

2025-12-19 · Tung-Ling Li, Yuhao Wu, Hongliang Liu arxiv

LLM-as-a-Judge systems supply the reward signal in modern RLHF and RLVR pipelines, but their binary verdict reduces to a single linear readout F_gap on one hidden state. We show this readout is shallow enough that short, low-perplexity tokens flip the verdict from "No" to "Yes". These tokens are sampled from the judge's own next-token distribution at the response position, with no manual seed set and no gradient-based optimization. Our procedure, AdvJudge-Zero, reaches $>$90% ensemble false-positive rate on 22 of 24 (model, dataset) cells across six Qwen, Llama, and Gemma judges, versus 54-72% for the prior curated 10-token benchmark, and the discovered surface transfers cross-format to a 70B scalar reward model. The same discovered pool enables a defense: a LoRA fine-tune stratified by a 9-class mechanism taxonomy hardens cross-family generalization where naive sampling on the same pool fails, with mechanism breadth rather than pool size carrying the gain. Under GRPO training, the hardened judge eliminates the reward-collapse failures (false-positive spikes and length collapse) we observe in the unhardened baseline on both MATH and GSM8K at ten seeds per condition. The discovered pool, the mechanism taxonomy, and per-prompt flip records will be released under responsible disclosure.

📄 PDF Abstract BibTeX arXiv:2512.17375

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Towards Understanding and Harnessing the Effect of Image Transformation in Adversarial Detection

2022-01-04 · Hui Liu, Bo Zhao, Yuefeng Peng, Weidong Li 외

Deep neural networks (DNNs) are threatened by adversarial examples. Adversarial detection, which distinguishes adversarial images from benign images, is fundamental for robust DNN-based services. Image transformation is …

Framing Instability in LLM Ethical Stance: Auditing Negation Sensitivity in Moral Dilemmas

2026-01-29 · Katherine Elkins, Jon Chun arxiv

Language models are increasingly consulted on ethically consequential questions, yet the stance a model expresses may not survive a change in framing. We audit 16 models across 14 ethically fraught dilemmas using polarit…

LLM Judges Can Be Too Generous When There Is No Reference Answer

2026-07-14 · Chalamalasetti Kranti, Sowmya Vajjala arxiv

LLM judges are increasingly being used to evaluate open-ended model responses, often in no-reference settings where a ground-truth answer is unavailable. However, can they reliably assess in such evaluation setups? We ex…

How Reliable Is Your Jailbreak Judge? Calibration and Adversarial Robustness of Automated ASR Scoring

2026-06-24 · Yang Gao arxiv

Almost every paper on LLM jailbreaks and prompt injection reports an attack-success rate (ASR), and that number is assigned not by people but by an automated judge: either a safety classifier trained for the task, or a g…

Adversarial Robustness

Does the Judge Prefer English? Evaluating Language-Switching Invariance in LLM-as-a-Judge

2026-06-12 · Shaojie Yin arxiv

Large language models (LLMs) are now widely used as automatic judges for open-ended instruction-following evaluation. This practice is convenient, scalable, and often more semantically aware than reference-based metrics,…