Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
TransPhy: Visual In-Context Learning for Physically Grounded Image EditingSenseTime · arXiv (Cornell University)Aug 25Preference Data Selection for Mitigating the Alignment Tax in Large Language ModelsMicrosoft · arXiv (Cornell University)Aug 25Native Multimodal Representation Learning for Click-Through Rate Prediction in E-Commerce ScenariosAlibaba · arXiv (Cornell University)Aug 25Maia 200: A Software Defined Dataflow System for Large-scale AI AccelerationMicrosoft · arXiv (Cornell University)Aug 25Contrastive Branch Policy OptimizationAlibaba · arXiv (Cornell University)Aug 25Rubrics as Visual-Repair Context for Self-Evolving UI-to-Code GenerationMicrosoft · arXiv (Cornell University)Aug 25Revisiting Structural Dependency in Autoregressive Multi-task Table Recognition via Order-Independent Cell-Level RepresentationsPreferred Networks · Lecture notes in computer scienceAug 25Demystifying Reinforcement Learning Post-Training of Language ModelsAllen Institute for AI · arXiv (Cornell University)Aug 24AffineTok: Semantic Affine Consistency for Diffusion-Friendly Visual TokenizerAlibaba · arXiv (Cornell University)Aug 24AI Finds A WayGoogle DeepMind · arXiv (Cornell University)Aug 24AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution TracesMicrosoft · arXiv (Cornell University)Aug 24TailSieve: Partial-Rollout-Guided Tail Routing for LLM RolloutsAlibaba · arXiv (Cornell University)Aug 24Beyond the Stability-Exploration Dilemma: Environmental Regularization for LLM Policy OptimizationAlibaba · arXiv (Cornell University)Aug 24From Inertia to Objectivity: Improving Deep Research Agents with Noise IsolationAlibaba · arXiv (Cornell University)Aug 24SACHA: Semantic-Aware Compression for 3D Gaussian Head AvatarsAlibaba · arXiv (Cornell University)Aug 24CaRGo-T: Causal Reasoning Graph-of-Thought improves Multimodal Humor ComprehensionMicrosoft · arXiv (Cornell University)Aug 24MobilePA-Bench: Benchmarking Mobile Planner Agents on Complex Real-World TasksAlibaba · arXiv (Cornell University)Aug 24WnW: Waxing-and-Waning KV Cache for Long-Form Speech LLMsAlibaba · arXiv (Cornell University)Aug 24What Process Evaluation of Coding Agents Actually Measures: Action, Task, and Step Are Three Different LevelsAlibaba · arXiv (Cornell University)Aug 24Make Lemonade out of Lemons: A Learning-based ZNS SSDs with Built-in CompressionAlibaba · ACM Transactions on StorageAug 24A \({(2+\varepsilon )}\)-Approximation Algorithm for Metric \({k}\)-MedianAlphabet · SIAM Journal on ComputingAug 24Can Large Language Models "Hyper-Thread"?Alibaba · arXiv (Cornell University)Aug 23ChainPrune: Evaluating and Reducing Redundancy in Long Chain-of-Thought ReasoningAlibaba · arXiv (Cornell University)Aug 22GuardPaint:SpeculativeSafetyDecodingforText-to-ImageGenerationMeta · arXiv (Cornell University)Aug 22PowerSlider: Exploiting Phase Asymmetry for LLM Serving under Demand ResponseMicrosoft · arXiv (Cornell University)Aug 22