Preliminary reweighting is established as a unifying and predictive framework for understanding and mitigating ICL jailbreak in MLLMs and a posterior-aware inference-time defense is introduced that adaptively injects benign counter-evidence based on estimated risk, effectively suppressing harmful posterior drift while...
A lightweight hierarchical attribution model is trained that predicts safety violations and attributes them to contributing user turns and token spans and preserves low false positive rates on benign conversations with high-risk vocabulary.
A methodology for mitigating sycophancy that employs the Bayesian Truth Serum (BTS), a peer-prediction mechanism, as the reward in Group Relative Policy Optimization (GRPO) to fine-tune an LLM.
LLMs have advanced rapidly, raising growing concerns about their safety. Recent work has proposed approaches to detect and defend against attacks including defenses at decoding stage that leverage models'hidden states. However, existing decoding-stage defenses suffer from two limitations. First, they introduce a trade-...
Euston is presented, an 8B mathematical claim-verification model trained to resist the all-false composition of the official evaluation sets and the low precision implied at realistic error prevalence, which is principally the all-false composition of the official evaluation sets.
A reference-based method that audits bias in hidden-state representations across related model variants across related model variants, for example before and after fine-tuning, which is complementary to output-based auditing rather than a replacement for it.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
