Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
Behavior2Value: Benchmarking and Empowering LLMs for Consumer Value Measurement from E-commerce BehaviorsAlibaba · arXiv (Cornell University)Sep 16A Zeroth-Order Paradigm for LLM Preference AlignmentAlibaba · arXiv (Cornell University)Sep 16DualSQL: Text-to-SQL with Multi-Agent Reinforcement LearningAlphabet · arXiv (Cornell University)Sep 16PULSE: Unlocking Practical Image Compression on Single-Thread CPUMicrosoft · arXiv (Cornell University)Sep 16LIGE-GR: A Smooth Leap from Ranking to Generative Recommendation in the LLM EraMeta · arXiv (Cornell University)Sep 16SVMemAgent: A Streaming Video Memory Agent for Query-Agnostic Online Frame SelectionMeta · arXiv (Cornell University)Sep 16ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE TasksMicrosoft · arXiv (Cornell University)Sep 16ScienceIDE: Turning World's Scientific Codebase into Agent Learnable EnvironmentsAlibaba · arXiv (Cornell University)Sep 16Ask the Tool, Don't Guess: Agent Tool Calls Hold Their Progress, and the Serving System Should Read ItAlibaba · arXiv (Cornell University)Sep 16Decodability is Not Causality: Dissociating Probe Readouts from Behavioral Drivers via SAE DecompositionAlphabet · arXiv (Cornell University)Sep 16HBFlex: A Flexible Memory System for Bridging Fine-Grained LLM States and Coarse-Grained HBF Parallel ExecutionAlibaba · arXiv (Cornell University)Sep 16GeoMesh: Workload-Balanced and Sign-Compressed Geo-Distributed LLM TrainingMicrosoft · arXiv (Cornell University)Sep 16Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM AgentsAlibaba · arXiv (Cornell University)Sep 16QuanText: Protecting Dataset-Level Secrets in Textual Data SharingMicrosoft · arXiv (Cornell University)Sep 16Hamming Ideals and Grobner Bases for ISD-like Syndrome DecodingTechnology Innovation Institute · arXiv (Cornell University)Sep 16WeaveRL: Weaving Reconstruction into Scene-Aware Fabrics for Perceptive Reinforcement LearningNVIDIA · arXiv (Cornell University)Sep 16An Iterative, Pathologist-in-the-Loop Workflow for Generation of Clinical-Grade Synthetic Pathology Images in a Diverse Cohort of Pancreatic TumorsMicrosoft · CancersSep 16Revealing the Pitfalls and Re-evaluating the Advancement of Heterophilic Graph LearningNVIDIA · Lecture notes in computer scienceSep 16GameReplica: A Benchmark for Black-Box Visual Game Replication by Vision-Language AgentsAlibaba · arXiv (Cornell University)Sep 15Locating Hidden Failures Makes Long-Horizon Agents More ReliableAlphabet, Google DeepMind · arXiv (Cornell University)Sep 15Tight Lower Bounds for Differentially Private Continual CountingAlphabet · arXiv (Cornell University)Sep 15Query-Aware Source-Risk Triage for Retrieval-Augmented GenerationAlphabet · arXiv (Cornell University)Sep 15G3AR: Graph-Guided Neural Visual Geometry for Scalable Multi-Sequence Aerial RegistrationNVIDIA · arXiv (Cornell University)Sep 15Turn-level Multiscale Density Ratio Estimation for LLM AgentsAlibaba · arXiv (Cornell University)Sep 15Verifiable Social Reasoning for LLM AssistantsAlphabet · arXiv (Cornell University)Sep 15