This paper studies the role of sinks and diagonal patterns as attention switch and anti-oversmoothing mechanisms. We analyze geometric conditions under which sinks can be represented, showing a necessary alignment between the embedding of the sink and all other embeddings. Next, we refine the current understanding of t...
Peter S\'uken\'ik, Cristina L\'opez Amado, Christoph H. Lampert et al.· 0 citations
Feedback-based adaptive quantum optimization (FALQON) is a promising approach for combinatorial optimization on noisy intermediate-scale quantum devices, but prior studies report that standard fixed-parameter schedules can require relatively deep circuits to reach strong solution quality. We study Optimal FALQON, a lay...
By turning Mathlib PR histories into a supervised signal, MathlibPR provides a step toward reviewer assistants and reward models that could help evaluate PRs and steer LLMs toward producing merge-ready Mathlib contributions.
Scientific diagrams help researchers communicate complex methods and relationships, but creating them requires substantial effort in content organization, layout design, and visual refinement. Automated diagram generation can reduce this workload. Reference examples can guide this process by showing how scientific idea...
Ling Yue, Jiaying Wang, Zhexi Lu et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Mixture-of-Experts (MoE) models have become the dominant architecture for large-scale language models, yet on-premises serving remains fundamentally memory-bound as batching turns sparse per-token compute into dense memory activation. Memory-centric architectures (PIM, NMP) improve bandwidth but leave compute underutil...
Yuseon Choi, Jingu Lee, Jungjun Oh et al.· 0 citations
This work proposes an evidence-driven keyframe sampling framework motivated by an information-theoretic perspective, and trains a query-conditioned evidence scoring network using annotated answer-supporting segments to motivate a tractable frame-wise selection criterion.
Yiheng Wang, Li-Chen Zhu, Yueqian Lin et al.· arXiv.org· 4 citations
With the rapid advancement of underwater net-working and multi-agent coordination technologies, autonomous underwater vehicle (AUV) ad-hoc networks have emerged as a pivotal framework for executing complex maritime missions, such as multi-target tracking. However, traditional data-centricarchitectures struggle to maint...
Kai Tian, Jialun Wang, Chuan Lin et al.· 0 citations
Surface electromyographic (sEMG) signals provide information about muscle activity and function during movement. Fatigue alters multiple properties of these signals, with patterns that vary across individuals and contraction intensities. However, physiological insights into fatigue remain underused in sEMG modeling, pa...
Tong Zhang, Hong Guo, Shuangzhou Yan et al.· 0 citations
When Vision-Language Models (VLMs) answer correctly, do they rely on the image? We introduce a Tri-Layer Diagnostic Framework that asks three questions of every (image, question) pair when the image is blanked: does the model register that the image is missing (Latent Anomaly Detection, LAD), does its output distributi...
Tool-using artificial intelligence (AI) agents are increasingly adopted as systems that plan, observe their environment, and take actions over extended time periods. This shift challenges the current evaluation practice, in which models are tested in restricted, fully observable settings. We argue that such evaluations...
Simone Aonzo, Merve Sahin, Aur\'elien Francillon et al.· 0 citations
Biological multimodal large language models (MLLMs) have emerged as powerful foundation models for scientific discovery. However, existing models are specialized to a single modality, limiting their ability to solve inherently cross-modal scientific problems. While model merging is an efficient method to combine the di...
Wonbin Lee, Dongki Kim, Sung Ju Hwang· 0 citations
Wearable and mobile sensing models infer health and well-being in everyday life, yet strong average performance can conceal failures for particular people, contexts, or devices. Existing fairness methods often require sensitive attributes during training, although such attributes may be unavailable, privacy-sensitive,...
Shaily Roy, Harshit Sharma, Daniel A. Adler et al.· 0 citations
