11 papers
How Faithful Is Trajectory-Based Data Attribution? Error Sources, Remedies, and Practical Guidelines
Junwei Deng, Pingbang Hu, Suliang Jin +4
Trajectory-based data attribution methods estimate the influence of training samples on model predictions by unrolling the training trajectory. They are widely used in applications…
Can Small Training Runs Reliably Guide Data Curation? Rethinking Proxy-Model Practice
Jiachen T. Wang, Tong Wu, Kaifeng Lyu +4
Data teams at frontier AI companies routinely train small proxy models to make critical decisions about pretraining data recipes for full-scale training runs. However, the communit…
Reward Hacking as Equilibrium under Finite Evaluation
Jiacheng Wang, Jinbin Huang
We prove that under five minimal axioms -- multi-dimensional quality, finite evaluation, effective optimization, resource finiteness, and combinatorial interaction -- any optimized…
Adversarial Déjà Vu: Jailbreak Dictionary Learning for Stronger Generalization to Unseen Attacks
Mahavir Dabas, Tran Huynh, Nikhil Reddy Billa +8
Large language models remain vulnerable to jailbreak attacks that bypass safety guardrails to elicit harmful outputs. Defending against novel jailbreaks represents a critical chall…
A Sustainable AI Economy Needs Data Deals That Work for Generators
Ruoxi Jia, Luis Oala, Wenjie Xiong +4
We argue that the machine learning value chain is structurally unsustainable due to an economic data processing inequality: each state in the data cycle from inputs to model weight…
Does More Inference-Time Compute Really Help Robustness?
Tong Wu, Chong Xiang, Jiachen T. Wang +4
Recently, Zaremba et al. demonstrated that increasing inference-time computation improves robustness in large proprietary reasoning LLMs. In this paper, we first show that smaller-…