CVPR 2026 accepted papers on Large language models
56 of the 4,042 papers accepted at CVPR 2026 (1.4%) are on Large language models, compared with 1.3% at CVPR 2025. They fall into 7 topics.
See CVPR 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Reasoning and chain of thought · 14 papers
- CARE What Fails: Contrastive Anchored-REflection for Verifiable Multimodal Reasoning
- From Scale to Speed: Adaptive Test-Time Scaling for Image Editing
- Reading or Reasoning? Format Decoupled Reinforcement Learning for Document OCR
- Reasoning Palette: Modulating Reasoning via Latent Contextualization for Controllable Exploration for (V)LMs
- Reinforcing Structured Chain-of-Thought for Video Understanding
- ReLaX: Reasoning with Latent Exploration for Large Reasoning Models
- Self-Consistency for LLM-Based Motion Trajectory Generation and Verification
- Stable and Efficient Single-Rollout RL for Multimodal Reasoning
- Test-Time Instance-Specific Parameter Composition: A New Paradigm for Adaptive Generative Modeling
- Towards Reasoning-Preserving Unlearning in Multimodal Large Language Models
- VisionLeaf: Entropy-Guided Leaf-First Reasoning for Efficient and Accurate Think-with-Image
- WPT: World-to-Policy Transfer via Online World Model Distillation
and 2 more in the interactive map.
Agents, code and math · 11 papers
- CME-CAD: Heterogeneous Collaborative Multi-Expert Reinforcement Learning for CAD Code Generation
- E-comIQ-ZH: A Human-Aligned Dataset and Benchmark for Fine-Grained Evaluation of E-commerce Posters with Chain-of-Thought
- GardenDesigner: Encoding Aesthetic Principles into Jiangnan Garden Construction via a Chain of Agents
- KnowVal: A Knowledge-Augmented and Value-Guided Autonomous Driving System
- Open-Ended Instruction Realization with LLM-Enabled Multi-Planner Scheduling in Autonomous Vehicles
- PSDesigner: Automated Graphic Design with a Human-Like Creative Workflow
- Universal Guideline-Driven Image Clustering via a Hybrid LLM Agent
- VIRO: Robust and Efficient Neuro-Symbolic Reasoning with Verification for Referring Expression Comprehension
- MCHDoc: A Comprehensive Benchmark for Reading Multi-Carrier Chinese Historical Documents
- OmniDocLayout: Towards Diverse Document Layout Generation via Coarse-to-Fine LLM Learning
- RAAS: LLM Agentic System Architecture Search with GRPO
Language, safety and interpretability · 11 papers
- Conditional Factuality Controlled LLMs with Generalization Certificates via Conformal Sampling
- DSO: Direct Steering Optimization for Bias Mitigation
- Interpretable and Steerable Concept Bottleneck Sparse Autoencoders
- Lyapunov Probes for Hallucination Detection in Large Foundation Models
- Pointer-CAD: Unifying B-Rep and Command Sequences via Pointer-based Edges & Faces Selection
- SelecTKD: Selective Token-Weighted Knowledge Distillation for LLMs
- Translating Signals to Languages for sEMG-Based Activity Recognition
- AudioStory: Generating Long-Form Narrative Audio with Large Language Models
- BoostSLT: Boosting Sign Language Translation via a Plug-and-Play Diffusion-Based Semantic Enhancer
- COPYLENS: Towards Copyrighted Characters Infringement Detection via Copyright-Aware Prompt Learning
- SAT-RRG: LLM-Guided Self-Adaptive Training for Radiology Report Generation with Token-Level Push-Pull Optimization
Instruction and fine-tuning · 7 papers
- Differences That Matter: Auditing Models for Capability Gap Discovery and Rectification
- Immunizing Models Against Harmful Long-Horizon Fine-Tuning via Contractive Optimization Dynamics
- SG-LoRA: Semantic-guided LoRA Parameters Generation
- Towards Efficient Medical Reasoning with Minimal Fine-Tuning Data
- TrafficAlign: Aligning Large Language Models for Traffic Scenario Generation
- Phrase-Grounding-Aware Supervised Fine-Tuning for Chart Recognition via Side-Masked Attention
- Vocabulary Scaling Law: Tuning Open-vocabulary Predictors for Their Openness
Alignment and preferences · 6 papers
- BeautyGRPO: Aesthetic Alignment for Face Retouching via Dynamic Path Guidance and Fine-Grained Preference Modeling
- SoliReward: Mitigating Susceptibility to Reward Hacking and Annotation Noise in Video Generation Reward Models
- Unified Generation and Self-Verification for Vision-Language Models via Advantage Decoupled Preference Optimization
- Circular-DPO: Aligning Multi-Stage 3D Generative Models via Preference Feedback Loop
- Phrase-grounded APO for Improving Chest X-ray Report Generation
- SMAP: Semantic Route Planning with Map-Grounded Multimodal Alignment
AI and society · 4 papers
- Benchmarking PhD-Level Coding in 3D Geometric Computer Vision
- GUIDE: A Benchmark for Understanding and Assisting Users in Open-Ended GUI Tasks
- Interactive Tracking: A Human-in-the-Loop Paradigm with Memory-Augmented Adaptation
- Neuro-Cognitive Reward Modeling for Human-Centered Autonomous Vehicle Control