VLMs Read Aloud: The Gap Between Recognizing and Using Visual Text
cs.CV, cs.CL
Submitted: 2026-11-01
Updated: 2026-11-03
Terminology
Sources
- Qwen3-VL Technical Report
- LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding
- Eliciting Latent Predictions from Transformers with the Tuned Lens
- PRISM-BENCH: Diagnosing Grounded and Faithful Multimodal Reasoning Beyond Final-Answer Accuracy
- Glyph: Scaling Context Windows via Visual-Text Compression
- ReadBench: Measuring the Dense Text Visual Reading Ability of Vision-Language Models
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- LongDocURL: a Comprehensive Multimodal Long Document Benchmark Integrating Understanding, Reasoning, and Locating
- Context Length Alone Hurts LLM Performance Despite Perfect Retrieval
- Reading is not Reasoning: Bridging the Agentic Policy Gap in Vision-Text Compression
- Look, Recite, Then Answer: Enhancing VLM Performance via Self-Generated Knowledge Hints
- IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations
- Decoupling semantics from vision: A framework for faithful visual-text compression evaluation
- Gemma 3 Technical Report
- Dissecting Recall of Factual Associations in Auto-Regressive Language Models
- GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning
- RULER: What's the Real Context Size of Your Long-Context Language Models?
- mPLUG-DocOwl 1.5: Unified Structure Learning for OCR-free Document Understanding
- Zoom Where the Answer Looks: Cross-Resolution Self-Distillation for MLLMs under Visual-Token Budgets
- LLMLingua: Compressing Prompts for Accelerated Inference of Large 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