AI 论文 · 2026-07
浏览 2026-07 发布的AI 论文内容，第 2 页，共 585 条。
- Shieldstral
- Explicit Layer Modeling for Video Object Insertion and Layer Decomposition
- DecoEvo: Score-Decoupled Co-Evolution of Solver and Rubric-Generator Skills in Text Space
- OmniDelta: Skill-Driven Budget Allocation for Token Compression in OmniLLMs
- CoRT: Counterfactual Replay for Token-Level Rubric-Guided Policy Optimization
- MemSFT: Mitigating Alignment Tax with an External Parametric Memory
- Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion
- ReDesign: Recovering Editable Design Structures from Images via Agentic Decomposition
- Visual prompt engineering for video models
- CodeNib: A Multi-View Data System for Serving Repository Context to Coding Agents
- SGTP: Sampling-based Game-Theoretic Planning for Real-Time Multi-Vehicle Autonomous Racing
- Memory for Large Language Models
- Temporal-Distance JEPA: Plan-Aware Representation Learning for Latent World Model Predictive Control
- Human-in-the-Loop Signature Bootstrapping for UAV Hyperspectral PFM-1 Mine Detection
- CAST: Game Solvers as Turn-Level Teachers for LLM Agents
- CLBench-V: Evaluating Multimodal Context Learning from Grounding to Knowledge Acquisition
- AMRD: Adaptive Multi-Teacher Relational Distillation for Lightweight Speech Emotion Recognition
- VisualPatchWorld: Code World Models as Latent Structured Representations for Planning
- OmniScope: Modality-Decoupled Token Compression for Omnimodal Large Language Models
- Lost in Compression: A Controlled Cross-Lingual Audit of Extractive Prompt Compressors
- OPERA: Offline Policy-guided Expert Routing and Adaptation for Universal Biomedical Image Analysis
- Towards Robust Reinforcement Learning for Small-Scale Language Model Agents
- PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models
- Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Model
- Agent Retrieval Bench: Evaluating Repository Context Retrieval for Coding Agents
- Data Pyramid for Embodied Manipulation
- ClinFusion: A Vision-Centric Multimodal LLM System for Holistic Medical Understanding
- Rethinking Classifier-Free Guidance in On-Policy Diffusion Distillation
- The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic Distillation
- Kimi K3: Open Frontier Intelligence
- Evidence Attribution in Visual Document Understanding without Coordinates or Region Labels
- DecoupleMix: Decoupled Ratio Search and Convex Allocation for Scalable VLM Data Recipes
- Keep It InMind: Benchmarking the Implicit-Association Blind Spot in Agent Memory
- From Proprietary to Open-Source: Bridging the Distribution Gap via Multi-Agent Protocol Distillation in Agentic Search
- FilmBench: A Film-Grade Benchmark for Cinematic Video Generation
- A New Role for Relevance: Guiding Corpus Interaction in Agentic Search
- Where Quality Breaks in Compressed Short-Text Generation: Staged Bottleneck Localization
- Grading the Narrators: An Isnad-Rijal Framework for Claim-Level Provenance in Multi-Agent Knowledge Systems
- Sol-Attn: Accelerating Video Generation Inference via On-the-Fly Attention Sparsification
- WorldDiT: A Unified Diffusion Architecture for World and Action Modeling
- GLI-AL: A Multi-Modal Glioma MRI Label Resource with Unified Anatomy-Lesion Labels
- OmniVAE: An Audio-Video VAE with Cross-Modal Alignment for Joint Generation
- DriveDNA: A Large-Scale Multimodal Naturalistic Driving Dataset and Benchmark for Driving Style Identification
- A Frozen 12B Beats Frontier Models on Verified Work: 100% Accuracy, 0 Tokens, Bit-Exact, Forever
- From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement
- N_0-TWAM: Scaling Tactile-Native World-Action Model for Contact-Rich Manipulation
- N_0-VTLA: Scaling Vision-Tactile-Language-Action Model with Latent Tactile Tokens
- Compute Globally, Materialize Locally: The Memory Contract of Sparse Event-KV
- GNM Head: A Generative aNthropometric Model of the human head
- JarvisHub: An Open Harness for Canvas-Native Multimodal Creative Agents
- Chamaileon: Cross-Context Binder Design with Contextualized Modeling and Mixed Sampling
- Novel Claim or Déjà Vu? Rethinking "Contamination-Free'' Dynamic Evaluation for Multimodal Automated Fact-Checking
- Characterizing Warp Divergence from Pascal to Blackwell
- When Activation Oracles Learn Not to Read: Concept-Specific Blind Spots in Fine-Tuned Oracles
- UltraViT: Latency-Optimized On-device Vision Encoder for Large Vision-Language Models
- Bitcoin Price Direction Prediction via Regime-Aware Multi-Modal Fusion of Social Sentiment and Technical Features
- IndicTalk: A Large-Scale Persona-Based Multilingual Conversational Corpus for Indic Languages
- dRAE: Representation Autoencoder with Hyper-Spherical Codes
- ID-V2V: Identity-Preserving Video Restylization
- StateAct: Program State, before Pixels, for Long-Horizon Computer-Use Agents
- Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills
- SceneActBench: Can Agents Act on the 3D Scenes They See?
- IDEAgent: Agentic Quality-Diversity Search for Research Idea Generation
- Projection Pursuit CPCANet for Domain Generalization
- Reasoning Denoiser: Denoising Reasoning Traces for Hallucination Detection in Large Reasoning Models
- Spectral Prior for Reducing Exposure Bias in Diffusion Models
- Scaling Native Multimodal Pre-Training From Scratch
- LAMAR: An Open Language-Aware Multilingual Alignment Reranker
- Leveraging External Knowledge for Historical Document Restoration via Retrieval-Augmented Large Language Models
- Closing the Loop: Training-Free Revisit Consistency for Autoregressive Generative Rendering
- What AI Red-Team Evaluations Can and Cannot Prove
- Oxygen-TryOn: Fashion-Native Foundation Model for Any-item Virtual Try-On
- Streaming Multi-Agent Autoregressive Diffusion Model with World State Registers
- GraphVid: Interactive Graph-Controllable Video Generation
- Self-Supervised Learning of Structured Dynamics from Videos
- OpenForgeRL: Train Harness-native Agents in Any Environment
- Visual Contrastive Self-Distillation
- SANA-Video 2.0: Hybrid Linear Attention with Attention Residuals for Efficient Video Generation
- Agentic Context Management: Solving Agent Memory and Cost by Treating Them as Lifecycle and Architecture Problems
- Recurrent Sinusoidal INRs for Efficient High-Fidelity Representation
- AREX: Towards a Recursively Self-Improving Agent for Deep Research
- ICAE-Bench: Evaluating Coding Agents as Interactive Project Builders
- Show, Don't Tell: Evaluating Spatial Cognition in Generative Pixels Rather Than LLM Text
- Sample-Efficient Learning from Agent Experience
- TableVerse: A Large-scale Tabletop Dataset with Real-world Grounded Layouts for Generalizable Manipulation
- Tencent WorkBuddy Bench: A Multi-Domain Coding-Agent Benchmark with Contamination-Resistant Task Construction
- Is Deep Research Reliable? Misleading Knowledge Induces False Conclusions
- K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs
- Super Star: Towards Streaming Real-time Interactive Agents for Digital Humans
- Are the Financial Reasoning from LLMs Credible? A Real World Test over Long-Horizon Statements
- Multi-Head Attention Residuals
- Progress Reward Modeling for Robotic Learning: A Comprehensive Survey
- Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning
- Robostral Navigate
- LLMs Get Lost in Evolving User Intent
- NVIDIA-labs OO Agents: Native Python Object-Oriented Agents
- ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion
- Train the Model, Not the Reader: Decodability Supervision for Verifiable Activation Explanations
- Self Gradient Forcing: Native Long Video Extrapolation
- SeededGrasp: Language-Guided Grasping in Complex Scenes with Multiple Embodiments