Papersby GlyphMicrosoft research papers
A company publishing mostly on agents and evaluation, best known here for LoRA.57 papers · 32,888 citations · latest Nov 2026
WebHarbor: Docking Real Websites for Evolving GUI Agent EnvironmentsWebHarbor provides realistic, containerized website environments for training and testing computer-using web agents across complex online tasks.
Microsoft28 upvotesRead →Agent World Model in the Wild: Computer-Use Agent Environment SynthesisThis pipeline creates controllable, open-source environments to safely train and test AI agents on complex, long-horizon computer-use tasks.Microsoft69 upvotesRead →Self-Evolving Coding Agents via Harness-Policy OptimizationThis framework improves coding agents by recursively alternating between optimizing the agent's policy and the environment harness it uses to interact with code.Microsoft48 upvotesRead →AesCode: Aesthetic Code Generation with Decoupled Cross-Modal RewardsThis paper introduces a method to improve code-based visual generation by using image-based aesthetic guidance and decoupled rewards to ensure quality and correctness.Microsoft137 upvotesRead →APT: Feedback-Driven Agentic Post-Training for Text-to-Image GenerationThis paper introduces an automated post-training framework that uses a feedback loop of evaluation and diagnosis to improve text-to-image models.Microsoft23 upvotesRead →LensErase: A Foundation Model for Erase AnythingThis paper introduces a foundation model designed for versatile image editing and object removal using mask-based inputs and efficient positional encoding.Microsoft52 upvotesRead →Simplifying and Improving On-Policy Distillation for Few-Step DiffusionThis paper introduces a simplified regression-based method for distilling diffusion models that improves generation speed without requiring complex training techniques.Microsoft27 upvotesRead →MiniCorp: The Last Mile of the AI Agent FirmMiniCorp is an office simulator that allows researchers to study how agents run companies and generate enterprise data for training.Microsoft39 upvotesRead →ActiveSaddler: Automated Curriculum Learning for Agent Harness OptimizationActiveSaddler optimizes agent training by dynamically adapting the curriculum of scenarios based on discovered failure patterns.Microsoft86 upvotesRead →Follow the Entities: A Corpus Map for Agentic SearchCorpusMap improves agentic search by organizing documents into a graph based on recurring entities, allowing agents to traverse relationships between files.Microsoft100 upvotesRead →Agensh: Scaling Organizational Intelligence to 1,024 AgentsAgensh is a scalable multi-agent harness that uses self-organized cooperation to coordinate up to 1,024 agents without a central orchestrator.Microsoft2 citations28 upvotesRead →The Tasteful Agent: Measuring and Improving Taste in Long-Horizon TasksTaste-Bench measures an agent's ability to make good long-horizon decisions by evaluating choices at critical decision forks in trajectories.Microsoft1 citations162 upvotesRead →When EOS Tokens Disagree: Understanding Length Inflation in On-Policy DistillationThis paper identifies termination-token mismatch as a cause of length inflation in policy distillation and provides a mitigation strategy.Microsoft6 citations109 upvotesRead →When2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning ModelsWhen2Think improves reasoning efficiency by dynamically allocating computation based on problem difficulty, using less compute for easy tasks.Microsoft1 citations54 upvotesRead →BI-Agent and BI-Bench: Towards Automating End-to-End Business IntelligenceBI-Agent automates end-to-end business intelligence tasks by decomposing workflows into subtasks like data search, join, and transformation.Microsoft1 citations30 upvotesRead →StudentSim: Training LLM-based Student SimulatorsStudentSim trains per-student simulators that answer like a given learner and change their answers under a tutor's guidance.Microsoft1 citations371 upvotesRead →AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution TracesAutoSaddler automatically improves LLM agent harnesses via offline failure-driven optimization, boosting performance on long-horizon benchmarks.Microsoft22 citations63 upvotesRead →Agent Lightning v1.0: Towards Harnessed Agentic RLAgent Lightning v1.0 enables reproducible reinforcement learning for arbitrary agent harnesses, substantially improving coding-agent performance with minimal data and compute.Microsoft11 citations35 upvotesRead →OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse PrefetchingOasisKV improves LLM inference throughput by storing full KV caches in lower memory tiers and prefetching only relevant entries into HBM using speculative-decoding lookahead predictions.Microsoft1 citations25 upvotesRead →Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation ModelMage-VL is a codec-native streaming vision-language model that uses motion-aware tokenization and a dual-system architecture to enable efficient real-time multimodal perception and reasoning.Microsoft17 citations36 upvotesRead →Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and EditingMage-Flow is a compact 4B generative stack combining a lightweight latent tokenizer and a native-resolution diffusion transformer to enable efficient high-resolution text-to-image generation and instruction-based editing.Microsoft4 citations77 upvotesRead →RESOURCE2SKILL: Distilling Executable Agent Skills from Human-Created Multimodal ResourcesRESOURCE2SKILL converts diverse multimedia tutorials and artifacts into structured, executable skills for software agents, improving performance through hierarchical retrieval, composition, and online acquisition.Microsoft3 citations141 upvotesRead →ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference OutcomesResearchStudio-Idea provides a skill suite for effective research ideation that combines literature search, novelty checking, and pattern-guided generation to produce traceable research proposals.Microsoft3 citations62 upvotesRead →ResearchStudio-Reel: Automate the Last Mile of Research from Paper to Poster, Video, and BlogResearchStudio-Reel automates research dissemination by composing specialized skills around a shared paper extractor, generating consistent and editable artifacts including posters, videos, and blogs with hard pass/fail quality gates.Microsoft2 citations60 upvotesRead →FastContext: Training Efficient Repository Explorer for Coding AgentsFastContext separates repository exploration from code solving in LLM agents using specialized exploration models that reduce token consumption and improve resolution rates.Microsoft5 citations94 upvotesRead →Latent Spatial Memory for Video World ModelsLatent spatial memory for video world models stores 3D scene information directly in diffusion latent space, eliminating pixel-space reconstruction overhead and achieving faster generation with reduced memory usage.Microsoft24 citations71 upvotesRead →WeaveBench: A Long-Horizon, Real-World Benchmark for Computer-Use Agents with Hybrid InterfacesWeaveBench presents a comprehensive benchmark for evaluating computer-use agents across multiple interfaces, revealing significant challenges in long-horizon task orchestration and highlighting the limitations of traditional performance assessment methods.Microsoft15 citations47 upvotesRead →Retrospective Harness Optimization: Improving LLM Agents via Self-Preference over Trajectory RolloutsRetrospective Harness Optimization (RHO) is a self-supervised method that improves AI agent performance by optimizing agent harness using only past trajectories through diverse task selection, parallel re-solving, and self-validation techniques.Microsoft29 citations36 upvotesRead →From Raw Experience to Skill Consumption: A Systematic Study of Model-Generated Agent SkillsLanguage agents benefit from reusable skills that encode domain-specific procedures, but their effectiveness varies significantly across different extraction and consumption scenarios, requiring careful evaluation and meta-skill guidance to optimize performance.Microsoft22 citations26 upvotesRead →SkillOpt: Executive Strategy for Self-Evolving Agent SkillsSkillOpt introduces a systematic text-space optimizer for agent skills that trains skills as external agent state with stable updates and zero deployment inference overhead, achieving superior performance across multiple benchmarks and execution environments.Microsoft173 citations263 upvotesRead →Lens: Rethinking Training Efficiency for Foundational Text-to-Image ModelsLens is a compact 3.8B-parameter text-to-image model achieving superior performance with reduced training compute through dense caption datasets, multi-resolution batching, efficient architecture, and optimization techniques.Microsoft5 citations103 upvotesRead →World-R1: Reinforcing 3D Constraints for Text-to-Video GenerationWorld-R1 framework improves video generation by incorporating 3D constraints through reinforcement learning and specialized text datasets while maintaining visual quality and scalability.Microsoft14 citations120 upvotesRead →Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?Self-distillation in large language models can degrade mathematical reasoning performance by suppressing uncertainty expression, particularly affecting out-of-distribution tasks.Microsoft110 citations57 upvotesRead →Online Experiential Learning for Language ModelsOnline Experiential Learning enables continuous improvement of language models through deployment experience by extracting and consolidating experiential knowledge via on-policy distillation.Microsoft28 citations61 upvotesRead →VIBEVOICE-ASR Technical ReportVibeVoice-ASR is a unified end-to-end speech understanding framework that processes long-form audio without chunking, supports multiple languages and code-switching, and uses prompt-based context injection for improved domain-specific accuracy.Microsoft31 citations25 upvotesRead →Proact-VL: A Proactive VideoLLM for Real-Time AI CompanionsProact-VL is a multimodal framework that enables real-time interactive AI companions for gaming scenarios with low-latency responses and strong video understanding capabilities.Microsoft19 citations37 upvotesRead →Exploratory Memory-Augmented LLM Agent via Hybrid On- and Off-Policy OptimizationEMPO² is a hybrid reinforcement learning framework that enhances exploration for large language model agents by integrating memory mechanisms with on- and off-policy updates, demonstrating improved performance and adaptability in complex environments.Microsoft22 citations36 upvotesRead →Experiential Reinforcement LearningExperiential Reinforcement Learning introduces an explicit experience-reflection-consolidation loop that improves learning efficiency and performance in sparse-reward environments by enabling structured behavioral revision without additional inference costs.Microsoft18 citations76 upvotesRead →CoPE-VideoLM: Codec Primitives For Efficient Video Language ModelsCoPE-VideoLM improves video language model efficiency by using video codec primitives instead of full-image encoding for most frames.Microsoft14 citations34 upvotesRead →Improving Data and Reward Design for Scientific Reasoning in Large Language ModelsA large-scale scientific question dataset and post-training pipeline are developed to improve open-ended science question answering through enhanced data processing and reinforcement learning with rubric-guided evaluation.Microsoft5 citations44 upvotesRead →Reinforcement World Model Learning for LLM-based AgentsReinforcement World Model Learning enables LLM-based agents to better anticipate action consequences and adapt to environment dynamics through self-supervised training that aligns simulated and real-world state transitions in embedding space.Microsoft24 citations28 upvotesRead →Self-Hinting Language Models Enhance Reinforcement LearningSAGE is an on-policy reinforcement learning framework that enhances GRPO by injecting self-hints during training to increase outcome diversity under sparse rewards, improving alignment of large language models.Microsoft21 citations31 upvotesRead →MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank RestorationTraining instability in large language models is linked to weight matrix stable rank decline and Jacobian alignment, which MSign addresses through matrix sign operations to prevent gradient explosions.Microsoft2 citations34 upvotesRead →Multiplex Thinking: Reasoning via Token-wise Branch-and-MergeMultiplex Thinking introduces a stochastic soft reasoning mechanism that samples multiple candidate tokens at each step to optimize reasoning trajectories with reinforcement learning while maintaining shorter sequences than traditional chain-of-thought methods.Microsoft19 citations40 upvotesRead →X-Coder: Advancing Competitive Programming with Fully Synthetic Tasks, Solutions, and TestsCode LLMs trained on fully synthetic data using a feature-based synthesis pipeline achieve superior performance on competitive programming benchmarks while reducing dependence on real-world coding datasets.Microsoft1 citations48 upvotesRead →DoVer: Intervention-Driven Auto Debugging for LLM Multi-Agent SystemsDoVer, an intervention-driven debugging framework, enhances reliability in LLM-based multi-agent systems by actively validating failure hypotheses and measuring task progress through targeted interventions.Microsoft37 citations28 upvotesRead →Black-Box On-Policy Distillation of Large Language ModelsGenerative Adversarial Distillation (GAD) enhances black-box distillation by framing the student model as a generator and using a discriminator to provide adaptive feedback, surpassing traditional sequence-level knowledge distillation.Microsoft63 citations54 upvotesRead →LoongRL:Reinforcement Learning for Advanced Reasoning over Long ContextsLoongRL, a data-driven reinforcement learning method, enhances long-context reasoning by transforming short multi-hop QA into high-difficulty tasks, improving accuracy and generalization in large language models.Microsoft34 citations63 upvotesRead →QueST: Incentivizing LLMs to Generate Difficult ProblemsQueST, a framework combining difficulty-aware graph sampling and fine-tuning, generates large-scale synthetic coding problems to enhance the performance of large language models in competitive coding and reasoning tasks.Microsoft6 citations36 upvotesRead →BitNet DistillationBitNet Distillation fine-tunes large language models to 1.58-bit precision using SubLN, multi-head attention distillation, and continual pre-training, achieving comparable performance with significant memory and inference speed improvements.Microsoft2 citations62 upvotesRead →
About and links
Topics: Agents and evaluation 13, Inference and efficiency 9, Training and scaling 7, Multimodal and robotics 7, Reasoning 7, Applied AI 4, Foundation models 4, Alignment and safety 3, Retrieval and data 3
Microsoft lab profile on Labs (labs.fru.dev): leaders, models, landmarksMicrosoft on Companies (companies.fru.dev)Also onAcquisitions·Quarterly·Paydays·Releases·TechConf