| Vision-as-a-Tool: Visual Harness for Multimodal Agentic CodingarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Interpretable weight-space programming builds transparent transformers and brain encoding modelsarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Scaling medical imaging report generation with multimodal reinforcement learningNature Communications · authors at Microsoft |
Microsoft |
Nov 2026 |
14 |
125 |
| Benchmarks Don't Talk Back: Learning Through Self-DisagreementarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Lazare, veni foras: Reprogramming Dormant Neurons to Backdoor AI AgentsarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| WordInferBench: Can Agents Infer How to Edit from Word Document Context?arXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Terms of Care: Designing for Participation in Data Governance by Disability CommunitiesarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Measuring Preferred Representation in AI-generated Images with Community ScorersarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Earth Embeddings in Environmental Remote SensingarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| EyeQ: Exploring Corneal Reflection for Mobile Context SensingarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| PARLEY: Evaluating LLM Collaboration Through Cooperative Asymmetric PuzzlesarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| A Theory of Policy Optimization in Post-Training: Coverage Collapse and Reference ResetsarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Tight Bounds for Sublinear Metric Steiner TreearXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| From Token Bills to Benchmark Weeks: Making Auto-Research for Data Systems Verifiable and AffordablearXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| FusionQ: A Generic and Optimal Decoding Framework for Ultra-Low-Latency Quantum Error CorrectionarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| Perturbing Inputs Improves Pretraining DistillationarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| LongCapHarness: Capability-Guided Auto-Harnessing for Long-Context ReasoningarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| WordEval: Evaluating Word-Native Operation Fidelity in Document EditingarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| When Experience Becomes a Backdoor: Poisoning Self-Evolving LLM Agents via Failure-Aware Task OptimizationarXiv · authors at Microsoft |
Microsoft |
Nov 2026 |
0 |
125 |
| HarnessMM: Recursive Self-Improvement for Harness Token Efficiencyauthors at Microsoft |
Microsoft |
Nov 2026 |
1 |
125 |
| WebHarbor: Docking Real Websites for Evolving GUI Agent Environmentslisted by Microsoft on Hugging Face |
Microsoft |
Nov 2026 |
0 |
125 |
| ParetoSaddler: Pareto-Guided Optimization of Agent Harnesses for Quality and Costlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| Mage-VL-Cowork: A Full Training Recipe to Learn Hybrid Computer Use From Evolved Harnesseslisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| ThinkBit: Search Codewords, Recalibrate Scales, Select by Reasoninglisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| Breaking Diffusion: Parallel Reasoning with Draft Refinementlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| OfficeTown: Generative Organizations for Evolving Agent Intelligencelisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| LensErase: A Foundation Model for Erase Anythinglisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| Simplifying and Improving On-Policy Distillation for Few-Step Diffusionlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewardslisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| AutoEnvScaling: Automating the Data Flywheel with Terminal Agentslisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| From Prompting to Composing: A Spatial Canvas Interface for Poster Generationlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| MiniCorp: The Last Mile of the AI Agent Firmlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| Capability-Driven Self-Evolution of Agent Memorylisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| LLM-as-Jev: LLMs Are Already Jev-Style Decision Models -- When and How to Fine-Tune Themlisted by Microsoft on Hugging Face |
Microsoft |
Oct 2026 |
0 |
125 |
| ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimizationlisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| Follow the Entities: A Corpus Map for Agentic Searchlisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| When2Think: Learning When and How Much to Reasonlisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| Rubric Rewards from Item Response Theorylisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| The Tasteful Agent: Measuring and Improving Taste in Long-Horizon Taskslisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillationlisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| VibeVoice-ASR-Streaming Technical Reportlisted by Microsoft on Hugging Face |
Microsoft |
Sep 2026 |
0 |
125 |
| Sliding-window beats linear attentionlisted by Microsoft on Hugging Face |
Microsoft |
Aug 2026 |
0 |
125 |
| AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution Traceslisted by Microsoft on Hugging Face |
Microsoft |
Aug 2026 |
0 |
125 |
| One Success Isn't Reliability: Thinkingbox, a Sandbox and Benchmark for Agents in Stateful Business Workflowslisted by Microsoft on Hugging Face |
Microsoft |
Aug 2026 |
0 |
125 |
| Agent Lightning v1.0: Towards Harnessed Agentic RLlisted by Microsoft on Hugging Face |
Microsoft |
Aug 2026 |
0 |
125 |
| OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse Prefetchinglisted by Microsoft on Hugging Face |
Microsoft |
Aug 2026 |
0 |
125 |
| AgentStream: How Well Do Self-Evolving LLM Agents Perform Under Streaming Tasks?listed by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Modellisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| OpenForgeRL: Train Harness-native Agents in Any Environmentlisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editinglisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| XL-DocBench: Benchmarking Evidence-Grounded Extra-Long Document Understandinglisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| DocAtlas: Long-Document Understanding as Mutable-State Interactionlisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| ResearchStudio-Reel: Automate the Last Mile of Research from Paper to Poster, Video, and Bloglisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference Outcomeslisted by Microsoft on Hugging Face |
Microsoft |
Jul 2026 |
0 |
125 |
| SeKV: Resolution-Adaptive KV Cache with Hierarchical Semantic Memory for Long-Context LLM Inferencelisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| HealthAgentBench: A Unified Benchmark Suite of Realistic Agentic Healthcare Environments for Challenging Frontier AI Agentslisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| Building to the Test: Coding Agents Deliver What You Check, Not What You Requestedlisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| Temporal Preference Optimization for Unsupervised Retrievallisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| FastContext: Training Efficient Repository Explorer for Coding Agentslisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| A Benchmark and Framework for Evaluating Next Action Predictions in Spreadsheetslisted by Microsoft on Hugging Face |
Microsoft |
Jun 2026 |
0 |
125 |
| Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?arXiv · authors at Microsoft |
Microsoft |
Mar 2026 |
113 |
125 |
| The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software DevelopersManagement Science · authors at Microsoft |
Microsoft |
Feb 2026 |
357 |
125 |
| On-Policy Context Distillation for Language ModelsarXiv · authors at Microsoft |
Microsoft |
Feb 2026 |
236 |
125 |
| Blueprint for fault-tolerant quantum computation with topological qubit arraysPhysical Review Research · authors at Microsoft |
Microsoft |
Dec 2025 |
106 |
125 |
| Native and Compact Structured Latents for 3D GenerationarXiv · authors at Microsoft |
Microsoft |
Dec 2025 |
189 |
125 |
| PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic MemoryarXiv · authors at Microsoft |
Microsoft |
Dec 2025 |
117 |
125 |
| ACON: Optimizing Context Compression for Long-horizon LLM AgentsarXiv · authors at Microsoft |
Microsoft |
Oct 2025 |
170 |
125 |
| Analog optical computer for AI inference and combinatorial optimizationNature · authors at Microsoft |
Microsoft |
Sep 2025 |
104 |
125 |
| Vibe coding: programming through conversation with artificial intelligencearXiv · authors at Microsoft |
Microsoft |
Jun 2025 |
146 |
125 |
| 'It was 80% me, 20% AI': Seeking Authenticity in Co-Writing with Large Language ModelsProceedings of the ACM on Human-Computer Interaction · authors at Microsoft |
Microsoft |
May 2025 |
111 |
125 |
| Agentic Reasoning and Tool Integration for LLMs via Reinforcement LearningarXiv · authors at Microsoft |
Microsoft |
Apr 2025 |
175 |
125 |
| MineWorld: a Real-Time and Open-Source Interactive World Model on MinecraftarXiv · authors at Microsoft |
Microsoft |
Apr 2025 |
126 |
125 |
| Dion: Distributed Orthonormalized UpdatesarXiv · authors at Microsoft |
Microsoft |
Apr 2025 |
109 |
125 |
| A Survey on Post-training of Large Language ModelsarXiv · authors at Microsoft |
Microsoft |
Mar 2025 |
143 |
125 |
| Toward an Evaluation Science for Generative AI SystemsarXiv · authors at Microsoft |
Microsoft |
Mar 2025 |
107 |
125 |
| Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement LearningarXiv · authors at Microsoft |
Microsoft |
Feb 2025 |
426 |
125 |
| Imagine while Reasoning in Space: Multimodal Visualization-of-ThoughtarXiv · authors at Microsoft |
Microsoft |
Jan 2025 |
270 |
125 |
| Can MLLMs Reason in Multimodality? EMMA: An Enhanced MultiModal ReAsoning BenchmarkarXiv · authors at Microsoft |
Microsoft |
Jan 2025 |
176 |
125 |
| ReFocus: Visual Editing as a Chain of Thought for Structured Image UnderstandingarXiv · authors at Microsoft |
Microsoft |
Jan 2025 |
106 |
125 |
| rStar-Math: Small LLMs Can Master Math Reasoning with Self-Evolved Deep ThinkingarXiv · authors at Microsoft |
Microsoft |
Jan 2025 |
426 |
125 |
| A Multimodal Biomedical Foundation Model Trained from Fifteen Million Image–Text PairsNEJM AI · authors at Microsoft |
Microsoft |
Dec 2024 |
535 |
125 |
| Phi-4 Technical ReportarXiv · authors at Microsoft |
Microsoft |
Dec 2024 |
1,918 |
125 |