Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
X-Ragent: Tool-Augmented Reinforcement Learning for Evidence-Grounded Chest X-Ray ReasoningAlibaba · arXiv (Cornell University)Oct 12REVEAL: Reverse Engineering Agentic Latents from Frontier-Model TracesAlibaba · arXiv (Cornell University)Oct 12Growing to See: Efficient Pretraining of Self-Supervised Vision TransformersAlphabet · arXiv (Cornell University)Oct 12Learning to GUIDE the Group: Self-Learning LLM Facilitation Across Multi-Party TasksAlphabet, Google DeepMind · arXiv (Cornell University)Oct 12HyWe: Unifying Search and Operation for Real-World Hybrid Web TasksAlibaba · arXiv (Cornell University)Oct 12GPU Goldmines: Tunable WebGPU Kernels for Edge LLM InferenceMicrosoft · arXiv (Cornell University)Oct 12HarnessTTT: Learn to Optimize Harness at Test-Time for Scientific DiscoveryNVIDIA · arXiv (Cornell University)Oct 12Tight End-to-End Accounting for Differentially Private Synthetic DataAlphabet · arXiv (Cornell University)Oct 12UniSGR-V2: Stepwise Target-Side Feature Anticipation and Crossing for Generative RecommendationAlibaba · arXiv (Cornell University)Oct 12FORGE: Procedural Weight Generation for Sparse MoE InferenceNVIDIA · arXiv (Cornell University)Oct 12Variable Input Masked Autoencoders for Wrist-Worn AccelerometryAlphabet · arXiv (Cornell University)Oct 12A Better-than-$2$ Approximation for $k$-Median Part II: Stable InstancesAlphabet · arXiv (Cornell University)Oct 12Architecting Instruction Execution for Scalable Fault-Tolerant Quantum ComputingMicrosoft · arXiv (Cornell University)Oct 12Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward MisalignmentAlibaba · arXiv (Cornell University)Oct 12LLM-HB: Language-Aware LLM-Guided Hotword Biasing for Code-Switching ASRiFlytek · arXiv (Cornell University)Oct 12Double-Rollout Reinforcement Learning with Sampling-based Reward EstimationNVIDIA · arXiv (Cornell University)Oct 12Recall Begets Recall: Recurrent Memory Ranking for Personalized LLMsAlibaba · arXiv (Cornell University)Oct 12Glasseek: A Transparent Recipe for Training and Trustworthy Evaluation of DeepSearch AgentsSenseTime · arXiv (Cornell University)Oct 12What Matters in Preference Post-Training of Continuous Autoregressive Speech Language ModelsMicrosoft · arXiv (Cornell University)Oct 12A Dual-Expert Routing Framework for Structured Data Extraction from Scientific TablesMicrosoft · ACM/IEEE Joint Conference on Digital LibrariesOct 12The Case for Single-Version Concurrency Control in the CloudMicrosoft · arXiv (Cornell University)Oct 11Scaling Repeated-Data Pretraining: Batch Size, Regularization, and GPU-Time TradeoffsAllen Institute for AI · arXiv (Cornell University)Oct 11Google Earth AI Geospatial Reasoning Agent: Capabilities and a Systematic EvaluationAlphabet · arXiv (Cornell University)Oct 11GA-SVD: Generation-aware Singular Value Decomposition for LLM CompressionTechnology Innovation Institute · arXiv (Cornell University)Oct 11What the Model Cannot See: Covert and Provable Text Invisibility in LLMsMicrosoft · arXiv (Cornell University)Oct 11