Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
V-JEPA Hand: Simple, Accurate and Fast 3D Hand and Camera Trajectory Recovery from Egocentric VideosMicrosoft · arXiv (Cornell University)Oct 19Symbolic Constraint Agents: Mining Formal Tool-Call Constraints to Prevent Agent ErrorsMicrosoft · arXiv (Cornell University)Oct 18DeGMatch: Target-Aware Reasoning Data Selection via Deconfounded Gradient MatchingLG AI Research · arXiv (Cornell University)Oct 18No Escape from the Quadratic Barrier: Adaptive Attacks on Linear $\ell_2$ SketchesAlphabet · arXiv (Cornell University)Oct 18Improving the Expressive Power of Hybrid Sequence Models via Multi-Hop RetrievalNVIDIA · arXiv (Cornell University)Oct 18Conjugate Compilation: Self-Improving Agentic Co-Compilation for Speculative DecodingMicrosoft · arXiv (Cornell University)Oct 18UnexpectedParrots: Iterating Behavioral Experiment Design with In-Silico SimulationsAllen Institute for AI · arXiv (Cornell University)Oct 18Epi: Towards a Persistent, Self-Improving Agent for Autonomous ResearchCohere, Microsoft · arXiv (Cornell University)Oct 18CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?SenseTime · arXiv (Cornell University)Oct 18Data.Readme: Documenting Table Relationships for Diverse Data TasksMicrosoft · arXiv (Cornell University)Oct 18Learning When to Abort in Transformer Planning: A Mechanistic StudyMicrosoft · arXiv (Cornell University)Oct 18Uber-Join: Unifying Diverse Table Joins using LLMs with Precision and Cost GuaranteesMicrosoft · arXiv (Cornell University)Oct 18Near-Optimal Last-Iterate Convergence of SGD in the Smooth Interpolation RegimeAlphabet · arXiv (Cornell University)Oct 18RARE: Reinforcement Learning with Adaptive Rubric Evolution for Open-Ended GenerationAlibaba · arXiv (Cornell University)Oct 18DuoLM: Fast and Effective Personalized Recommendation with Language ModelsGoogle DeepMind · arXiv (Cornell University)Oct 17Blackbox Performance Analysis of LLM-based Network ControlMicrosoft · arXiv (Cornell University)Oct 17Write Once, Query Many: Evolving Compact Reusable Memory for Long VideosMicrosoft · arXiv (Cornell University)Oct 17fxBench: Evaluating and Understanding Formula Suggestions in SpreadsheetsMicrosoft · arXiv (Cornell University)Oct 17Koi: Evidence-Guided, Self-Calibrating Planning for Multi-Cloud, Heterogeneous LLM Inference FleetsNVIDIA · arXiv (Cornell University)Oct 17Do Conversational Agents Need Explicit Memory? A Controlled Comparison with Raw-History RetrievalAlphabet · arXiv (Cornell University)Oct 17Do Recipes Have Personas? Characterizing and Generating Creator Style in Attributed Procedural GraphsMicrosoft · Lecture notes in computer scienceOct 17SPIDES: Information-Flow Control for Browser AgentsMicrosoft · arXiv (Cornell University)Oct 16Encoder-Free Multimodal LLMs Learn Implicit EncodersNVIDIA · arXiv (Cornell University)Oct 16Learning Shallow Quantum CircuitsAlphabet · SIAM Journal on ComputingOct 16Leveraging verifiable structure in quantum circuits to find more efficient compilations with agentic AIAlphabet · arXiv (Cornell University)Oct 16