13 papers
SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agents
Rui Melo, Riccardo Fogliato, Sean Zhou +2
Large language models (LLMs) are increasingly deployed in automated code-review systems, where their approvals can determine which code is merged into shared repositories. However,…
What Fits (Into Few Tokens) Doesn't Overfit: Compression and Generalization in ML Research Agents
Martin Andres Bertran, Aaron Roth, Zhiwei Steven Wu
Reusing a held-out benchmark adaptively should, in principle, invite overfitting. Yet benchmark-driven machine learning (ML) has produced surprisingly little overfitting in practic…
Back to Blackwell: Closing the Loop on Intransitivity in Multi-Objective Preference Fine-Tuning
Jiahao Zhang, Lujing Zhang, Keltin Grimes +3
A recurring challenge in preference fine-tuning (PFT) is handling (i.e., cyclic) preferences. Intransitive preferences often stem from either …
From Curiosity to Caution: Mitigating Reward Hacking for Best-of-N with Pessimism
Zhuohao Yu, Zhiwei Steven Wu, Adam Block
Inference-time compute scaling has emerged as a powerful paradigm for improving language model performance on a wide range of tasks, but the question of how best to use the additio…
Many AI Analysts, One Dataset: Navigating the Agentic Data Science Multiverse
Martin Bertran, Riccardo Fogliato, Zhiwei Steven Wu
Empirical conclusions depend not only on data but on analytic decisions made throughout the research process. Many-analyst studies have quantified this dependence: independent team…
Gained in Translation: Privileged Pairwise Judges Enhance Multilingual Reasoning
Lintang Sutawika, Gokul Swamy, Zhiwei Steven Wu +1
When asked a question in a language less seen in its training data, current reasoning large language models (RLMs) often exhibit dramatically lower performance than when asked the…