MATH-PT: The First Mathematical Reasoning Benchmark for Portuguese Unveiled
A research team has released the MATH-PT dataset, a mathematical reasoning evaluation benchmark specifically designed fo…
🏷️ AR
3347 articles about 'AR'
SpecTr-GBV: Multi-Draft Block Verification Accelerates Speculative Decoding
A new study proposes SpecTr-GBV, the first method to combine multi-draft strategies with block verification techniques, …
Breaking Defenses Word by Word: ICD Jailbreak Strategy Exposes New LLM Security Vulnerabilities
Researchers propose Incremental Completion Decomposition (ICD), a jailbreak strategy that guides large language models t…
Evaluation Results Released for RAG-Powered Virtual Assistant in Undergraduate Project Supervision
A new study explores the effectiveness of a generative AI virtual assistant based on Retrieval-Augmented Generation (RAG…
IBM Launches Granite 4.1: 8 Billion Parameter Model Rivals 32B MoE Models
IBM has officially released the Granite 4.1 series, in which the 8B dense model matches or even surpasses the performanc…
Unsupervised Graph Neural Network Modeling: A New Framework for Accounting Account Anomaly Detection
A latest arXiv paper proposes an unsupervised discriminative framework based on graph neural networks that abstracts acc…
DORA System: Asynchronous Reinforcement Learning Breaks Through LLM Training Bottlenecks
A research team has proposed the DORA system, a scalable asynchronous reinforcement learning architecture that addresses…
SWAN: Environment-Aware Adaptive Multimodal Networks
A research team has proposed the SWAN framework, enabling multimodal deep neural networks to dynamically adapt at runtim…
Entropy Centroids as Intrinsic Rewards: A New Paradigm for Test-Time Compute Scaling
A latest arXiv paper proposes the "Entropy Centroids" method, which scales LLM computation at test time without external…
New Causal Bandit Framework: Intelligent Decision-Making Under Budget Constraints
A latest arXiv paper proposes a Budget-Constrained Causal Bandit framework that bridges uplift modeling with sequential …
CNN, Transformer, and Mamba: Which Architecture Best Suits PPG-Based Emotion Recognition?
A new study systematically compares three deep learning architectures — CNN, Transformer, and Mamba — on photoplethysmog…
Privacy-Preserving Federated Learning Framework Empowers Distributed Chemical Process Optimization
A latest arXiv paper proposes a privacy-preserving federated learning framework for distributed chemical process optimiz…