Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
An Efficient Hybrid Sparse Attention with CPU-GPU Parallelism for Long-Context InferenceMicrosoft · arXiv (Cornell University)May 8LLMs Improving LLMs: Agentic Discovery for Test-Time ScalingAlphabet · arXiv (Cornell University)May 8MemCompiler: Compile, Don't Inject -- State-Conditioned Memory for Embodied AgentsMicrosoft · arXiv (Cornell University)May 8AdaTKG: Adaptive Memory for Temporal Knowledge Graph ReasoningLG AI Research · arXiv (Cornell University)May 8When Are Experts Misrouted? Counterfactual Routing Analysis in Mixture-of-Experts Language ModelsMicrosoft · arXiv (Cornell University)May 8Sample Complexity of Stochastic Optimization with Integer VariablesMicrosoft · arXiv (Cornell University)May 8PaT: Planning-after-Trial for Efficient Test-Time Code GenerationMicrosoft · arXiv (Cornell University)May 8Toward Better Geometric Representations for Molecule Generative ModelsAllen Institute for AI · arXiv (Cornell University)May 8Switchcraft: AI Model Router for Agentic Tool CallingMicrosoft · arXiv (Cornell University)May 8TeamBench: Evaluating Agent Coordination under Enforced Role SeparationAlphabet, Google DeepMind · arXiv (Cornell University)May 8Revisiting Transformer Layer Parameterization Through Causal Energy MinimizationMicrosoft · arXiv (Cornell University)May 8Bridging Textual Profiles and Latent User Embeddings for PersonalizationAlphabet · arXiv (Cornell University)May 7Delulu: A Verified Multi-Lingual Benchmark for Code Hallucination Detection in Fill-in-the-Middle TasksMicrosoft · arXiv (Cornell University)May 7Gated QKAN-FWP: Scalable Quantum-inspired Sequence LearningNVIDIA · arXiv (Cornell University)May 7PACEvolve++: Improving Test-time Learning for Evolutionary Search AgentsGoogle DeepMind · arXiv (Cornell University)May 7Mid-Circuit Measurements for Clifford Noise Reduction in Hamiltonian SimulationsNVIDIA · arXiv (Cornell University)May 7TrajGANR: Trajectory-Centric Urban Multimodal Learning via Geospatially Aligned Neural RepresentationsAlphabet · arXiv (Cornell University)May 7Regulating Branch Parallelism in LLM ServingNVIDIA · arXiv (Cornell University)May 7Uneven Evolution of Cognition Across Generations of Generative AI ModelsAlphabet · arXiv (Cornell University)May 7Accelerated Relax-and-Round for Concave Coverage ProblemsAlphabet · arXiv (Cornell University)May 7Agentic Coding Needs Proactivity, Not Just AutonomyAlphabet · arXiv (Cornell University)May 7Continuous First, Discrete Later: VQ-VAEs Without Dimensional CollapseAlphabet · arXiv (Cornell University)May 7Utilizing Pre-Trained Language Models and Large Language Models for 10-K Items SegmentationNVIDIA · Journal of Information SystemsMay 7ReasonSTL: Bridging Natural Language and Signal Temporal Logic via Tool-Augmented Process-Rewarded LearningAlibaba · arXiv (Cornell University)May 7XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural SensitivityMicrosoft · arXiv (Cornell University)May 7