SPECTRA is proposed, a training-free, weight-level method that restores refusal behavior by transferring the base model's refusal direction into the fine-tuned model's weights via an activation-aware, low-rank update, requiring no gradient-based optimization or domain-specific safety data.
V. Chhabra, Ding Zhu, Zhi-Hui Zhu et al.· 0 citations
Generative models are increasingly trained in self-consuming iterative loops, where users curate preferred samples from model-generated candidates and the curated samples are used to train future generations of the model. Prior work has largely assumed fixed user preferences, but in practice exposure to model outputs g...
Xiu-Kun Wei, Tian Xie, Ding Zhu et al.· 1 citation
SR-Fraud is presented, an outcome-supervised reflective LLM framework that decouples request-time decisions from offline adaptation, and improves all detection metrics over its frozen decision agent, obtains higher point estimates than static and periodically retrained CatBoost, and detects an emerging fraud burst.
Xu-Wei Tan, Yao Ma, Xue-Ru Zhang· arXiv.org· 1 citation
This paper develops theoretical bounds on the unfairness and expected loss at deployment, and derives sufficient conditions under which fairness and accuracy can be perfectly transferred via invariant representation learning, and designs a learning algorithm such that fair ML models learned with training data still hav...
A systematic study of how pruning affects SAE behavior is presented and theoretically shows that, for a fixed SAE, its impact is governed by perturbation energy, a covariance-weighted norm.
Suchit Gupte, Xue-Ru Zhang, M. Khalili· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.