Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
Self-drafting with Uno on a mixture of experts: serving and training an adapter for Gemma 4 26B A4BCerebras Systems · arXiv (Cornell University)Sep 3What Matters for Aggressive Decoding-Time KV Eviction? Temporal Aggregation and Ranking PreservationAlibaba · arXiv (Cornell University)Sep 3CulturalMenuBench: Probing the Knowledge-Application Gap in Multimodal Culinary ReasoningAlibaba · arXiv (Cornell University)Sep 3PPO-STGNN: A Proximal Policy Optimization Approach with Spatio-Temporal Graph Neural Networks for DAG Task Scheduling in Cloud-Edge-End ComputingAlphabet · arXiv (Cornell University)Sep 3On the geometry and typicality of quantum magicTechnology Innovation Institute · arXiv (Cornell University)Sep 3CrossViewDiff: A cross-view diffusion model for satellite-to-street view synthesisSenseTime · Pattern RecognitionSep 3ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action ModelsMicrosoft · arXiv (Cornell University)Sep 2VibeVoice-ASR-Streaming Technical ReportMicrosoft · arXiv (Cornell University)Sep 2Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiersMicrosoft · arXiv (Cornell University)Sep 2MemoryLACE: Memory Lifecycle-Aware Consolidation and Evidence RetrievalCerebras Systems · arXiv (Cornell University)Sep 2Triage: Boosting Cross-region Video Conferencing with Proactive FEC on Overlay NetworkAlibaba · Proceedings of the ACM on NetworkingSep 2CRAW: Codec Robust Audio WatermarkingNVIDIA · arXiv (Cornell University)Sep 2CHIME: Credit-Aware Hierarchical Memory Evolution for Long-Horizon Agentic PlanningAlibaba · arXiv (Cornell University)Sep 2ShikumiMiner: Mining Recurring Implementation Patterns in AI CodebasesMicrosoft · arXiv (Cornell University)Sep 2POI-Loc: A Fine-Grained POI Localization Benchmark and an Asymmetric Global-to-Local Matching MethodAlibaba · arXiv (Cornell University)Sep 2Overcoming the Randomness-Utility Trade-off in Answering Differentially Private Linear QueriesAlphabet · arXiv (Cornell University)Sep 2SPAR: Enhancing Industrial-Scale Generative POI Recommendation via Real-World Spatial PerceptionAlibaba · arXiv (Cornell University)Sep 2Diagnosing with Insights: Structured Analysis of Agent Failures via Behavioral AbstractionsMicrosoft · arXiv (Cornell University)Sep 2Decoupling Disaggregated Memory Optimizations from Indexing: A Compiler-Runtime ApproachMicrosoft · arXiv (Cornell University)Sep 2Exponential speedup of polarization stabilization for long distance DWDM quantum networksTechnology Innovation Institute · arXiv (Cornell University)Sep 2FairCache : Demystifying Cache-Induced Unfairness in Multi-Tenant Large Language Model ServingAlibaba · IEEE Transactions on Parallel and Distributed SystemsSep 2Output Type Before Quality: A Standards-Derived XAI Admissibility Rubric for Autonomous-Driving SafetyNVIDIA · Lecture notes in computer scienceSep 2Fin-PRM: A Domain-Specialized Process Reward Model for Financial Reasoning in Large Language ModelsAlibaba · International Joint Conference on Artificial Intelligence2 citationsSep 1CollabLLM: From Passive Responders to Active Collaborators (Extended Abstract)Microsoft · International Joint Conference on Artificial Intelligence1 citationSep 1AI-enhanced adaptive virtual screening of large libraries for ligand discoveryAlphabet, NVIDIA · Nature Biotechnology1 citationSep 1