Perturb, Repair, Verify: Self-Play Vision-Language Verifiers for Compositional Understanding
cs.CV, cs.LG
Submitted: 2026-11-03
Updated: 2026-11-03
Terminology
Sources
- The devil is in the fine-grained details: Evaluating open-vocabulary object detectors for fine-grained understanding
- Investigating the Role of Attribute Context in Vision-Language Models for Object Recognition and Detection
- ColorSwap: A Color and Word Order Dataset for Multimodal Evaluation
- ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation
- How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites
- InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning
- Dense and Aligned Captions (DAC) Promote Compositional Reasoning in VL Models
- HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Models
- VisPlay: Self-Evolving Vision-Language Models from Images
- SugarCrepe: Fixing Hackable Benchmarks for Vision-Language Compositionality
- R-Zero: Self-Evolving Reasoning LLM from Zero Data
- AI safety via debate
- A Visual Leap in CLIP Compositionality Reasoning through Generation of Counterfactual Sets
- RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback
- Enhancing Vision-Language Compositional Understanding with Multimodal Synthetic Data
- VLFeedback: A Large-Scale AI Feedback Dataset for Large Vision-Language Models Alignment
- Evaluating Object Hallucination in Large Vision-Language Models
- Let's Verify Step by Step
- Self-Improving VLM Judges Without Human Annotations
- Revisiting the Role of Language Priors in Vision-Language Models
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs