相关论文: Making Models Unmergeable via Scaling-Sensitive Lo…

Safety alignment in Large Language Models (LLMs) often creates a systematic discrepancy between a model's aligned output and the underlying pre-aligned data distribution. We propose a framework in which the effect of safety alignment on…

Model merging has attracted attention as an effective path toward multi-task adaptation by integrating knowledge from multiple task-specific models. Among existing approaches, dynamic merging mitigates performance degradation caused by…

To counter man-at-the-end attacks such as reverse engineering and tampering, software is often protected with techniques that require support modules to be linked into the application. It is well-known, however, that attackers can exploit…

The theory of state tracking in recurrent architectures has predominantly focused on expressive capacity: whether a fixed architecture can theoretically realize a set of symbolic transition rules. We argue that equally important is error…

Ensemble of models is well known to improve single model performance. We present a novel ensembling technique coined MAC that is designed to find the optimal function for combining models while remaining invariant to the number of…

Achieving backward compatibility when rolling out new models can highly reduce costs or even bypass feature re-encoding of existing gallery images for in-production visual retrieval systems. Previous related works usually leverage losses…

Deep neural networks are vulnerable to backdoor attacks. Among the existing backdoor defense methods, trigger reverse engineering based approaches, which reconstruct the backdoor triggers via optimizations, are the most versatile and…

Deep neural networks are vulnerable to backdoor attacks, where malicious behaviors are implanted during training. While existing defenses can effectively purify compromised models, they typically require labeled data or specific training…

Large language models (LLMs), despite being safety-aligned, exhibit brittle refusal behaviors that can be circumvented by simple linguistic changes. As tense jailbreaking demonstrates that models refusing harmful requests often comply when…

Adaptive model predictive control (MPC) robustly ensures safety while reducing uncertainty during operation. In this paper, a distributed version is proposed to deal with network systems featuring multiple agents and limited communication.…

Merging various task-specific Transformer-based models trained on different tasks into a single unified model can execute all the tasks concurrently. Previous methods, exemplified by task arithmetic, have been proven to be both effective…

Conformal prediction is a popular framework of uncertainty quantification that constructs prediction sets with coverage guarantees. To uphold the exchangeability assumption, many conformal prediction methods necessitate an additional…

Existing semantic communication schemes primarily focus on single-hop scenarios, overlooking the challenges of multi-hop wireless image transmission. As semantic communication is inherently lossy, distortion accumulates over multiple hops,…

Multi-agent systems powered by large language models are advancing rapidly, yet the tension between mutual trust and security remains underexplored. We introduce and empirically validate the Trust-Vulnerability Paradox (TVP): increasing…

Recent advances in Large Language Models (LLMs) have led to impressive alignment where models learn to distinguish harmful from harmless queries through supervised finetuning (SFT) and reinforcement learning from human feedback (RLHF). In…

As large language models (LLMs) become increasingly capable, it is prudent to assess whether safety measures remain effective even if LLMs intentionally try to bypass them. Previous work introduced control evaluations, an adversarial…

The rise of Large Language Models (LLMs) has heightened concerns about the misuse of AI-generated text, making watermarking a promising solution. Mainstream watermarking schemes for LLMs fall into two categories: logits-based and…

State machine formalisms equipped with hierarchy and parallelism allow to compactly model complex system behaviours. Such models can then be transformed into executable code or inputs for model-based testing and verification techniques.…

We study empirical scaling laws for language model merging measured by cross-entropy. Despite its wide practical use, merging lacks a quantitative rule that predicts returns as we add experts or scale the model size. We identify a compact…

Models trained on different datasets can be merged by a weighted-averaging of their parameters, but why does it work and when can it fail? Here, we connect the inaccuracy of weighted-averaging to mismatches in the gradients and propose a…