Safety Alignment Should Be Made More Than Just A Few Attention Heads arXiv:2508.19697v1 Announce Type: cross Abstract: Current safety alignment for large language models(LLMs) continues to present vulnerabilities, given that adversarial prompting can effectively bypass their safety measures.Our investigation shows…
Model Science: getting serious about verification, explanation and control of AI systems arXiv:2508.20040v1 Announce Type: new Abstract: The growing adoption of foundation models calls for a paradigm shift from Data Science to Model Science. Unlike data-centric approaches, Model Science places…
PSO-Merging: Merging Models Based on Particle Swarm Optimization arXiv:2508.19839v1 Announce Type: cross Abstract: Model merging has emerged as an efficient strategy for constructing multitask models by integrating the strengths of multiple available expert models, thereby reducing the need to fine-tune…
MovieCORE: COgnitive REasoning in Movies arXiv:2508.19026v1 Announce Type: cross Abstract: This paper introduces MovieCORE, a novel video question answering (VQA) dataset designed to probe deeper cognitive understanding of movie content. Unlike existing datasets that focus on surface-level comprehension, MovieCORE emphasizes…
Experimental End-to-End Optimization of Directly Modulated Laser-based IM/DD Transmission arXiv:2508.19910v1 Announce Type: cross Abstract: Directly modulated lasers (DMLs) are an attractive technology for short-reach intensity modulation and direct detection communication systems. However, their complex nonlinear dynamics make the modeling and…
Conditional Wasserstein Distances with Applications in Bayesian OT Flow Matching arXiv:2403.18705v3 Announce Type: replace Abstract: In inverse problems, many conditional generative models approximate the posterior measure by minimizing a distance between the joint measure and its learned approximation. While this…
Preference Elicitation for Multi-objective Combinatorial Optimization with Active Learning and Maximum Likelihood Estimation arXiv:2503.11435v2 Announce Type: replace-cross Abstract: Real-life combinatorial optimization problems often involve several conflicting objectives, such as price, product quality and sustainability. A computationally-efficient way to tackle multiple…
Revisiting Pre-trained Language Models for Vulnerability Detection arXiv:2507.16887v2 Announce Type: replace-cross Abstract: The rapid advancement of pre-trained language models (PLMs) has demonstrated promising results for various code-related tasks. However, their effectiveness in detecting real-world vulnerabilities remains a critical challenge. %…
Apple Intelligence Foundation Language Models: Tech Report 2025 arXiv:2507.13575v3 Announce Type: replace Abstract: We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon…
Emotions as Ambiguity-aware Ordinal Representations arXiv:2508.19193v2 Announce Type: replace Abstract: Emotions are inherently ambiguous and dynamic phenomena, yet existing continuous emotion recognition approaches either ignore their ambiguity or treat ambiguity as an independent and static variable over time. Motivated by…
