Multimodal LLMs reliably solve many CAPTCHA tasks but can be defended by adding fine-grained localization and implicit counting that drops state-of-the-art success from over 95% to 0%.
Symbols of One-Loop Integrals From Mixed Tate Motives
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
abstract
We use a result on mixed Tate motives due to Goncharov (arXiv:alg-geom/9601021) to show that the symbol of an arbitrary one-loop 2m-gon integral in 2m dimensions may be read off directly from its Feynman parameterization. The algorithm proceeds via recursion in m seeded by the well-known box integrals in four dimensions. As a simple application of this method we write down the symbol of a three-mass hexagon integral in six dimensions.
citation-role summary
background 3
citation-polarity summary
roles
background 3representative citing papers
Large-scale review mining of 1M+ comments from 171 Gen-AI apps using an LLM framework reveals top topics plus three opportunities and three challenges for developers.
COMPASS formalizes HPC configuration questions as ML tasks on traces, quantifies recommendation trustworthiness, and delivers 65.93% lower average job turnaround time plus 80.93% lower node usage versus prior methods in simulator tests.
The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.
Preference-based prompting raises LLM adherence to object-oriented design principles in UML generation but leaves substantial output variance and model-specific differences intact.
Survey and interview study finds neurodivergent computing students prefer structured collaborative active learning with small teams and explicit roles.
The agentic web requires new normative infrastructure of laws, norms, and practices to allow user-delegated AI agents to access online properties without being blocked as malicious bots.
DECICE delivers an Integrated AI Scheduler with RNN prediction and a Digital Twin for energy-aware workload scheduling in Kubernetes and Slurm environments as part of a European project.
citing papers explorer
-
Understanding the Challenges and Opportunities of Generative AI Apps: An Empirical Study
Large-scale review mining of 1M+ comments from 171 Gen-AI apps using an LLM framework reveals top topics plus three opportunities and three challenges for developers.
-
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.
-
Reliability of Large Language Models for Design Synthesis: An Empirical Study of Variance, Prompt Sensitivity, and Method Scaffolding
Preference-based prompting raises LLM adherence to object-oriented design principles in UML generation but leaves substantial output variance and model-specific differences intact.
-
"I can't read your mind": A Study of Neurodivergent Computing Students' Experiences with Collaborative Active Learning
Survey and interview study finds neurodivergent computing students prefer structured collaborative active learning with small teams and explicit roles.
-
The Agentic Web Requires New Normative Infrastructure
The agentic web requires new normative infrastructure of laws, norms, and practices to allow user-delegated AI agents to access online properties without being blocked as malicious bots.
-
DECICE: AI-Driven Scheduling and Digital Twin Integration for the Cloud-HPC-Edge Compute Continuum
DECICE delivers an Integrated AI Scheduler with RNN prediction and a Digital Twin for energy-aware workload scheduling in Kubernetes and Slurm environments as part of a European project.