Do VLMs Read or Rewrite? On Transcription Faithfulness in Vision-Language Models

arXiv:2607.21617 · cs.AI, cs.CL · Submitted 2026-05-29 · Read on arXiv

cs.AI, cs.CL

Submitted: 2026-05-29

Updated: 2026-09-02

Code: https://github.com/mindee/doctr

Terminology

• Benchmark — 17× in this paper · explained in 2 episodes
• Hallucination — 12× in this paper · explained in 3 episodes
• Adam — 4× in this paper · explained in 2 episodes
• Accuracy — 3× in this paper · explained in 1 episode
• Optical Character Recognition — 3× in this paper
• Transformer — 3× in this paper · explained in 2 episodes
• Word Error Rate — 3× in this paper
• Ablation — 2× in this paper · explained in 1 episode
• Cosine Similarity — 2× in this paper · explained in 3 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Large Language Model — 2× in this paper
• Robustness — 2× in this paper · explained in 5 episodes
• Ablation Studies — 1× in this paper · explained in 1 episode
• Adaptation — 1× in this paper · explained in 1 episode
• Adversarial Attacks — 1× in this paper · explained in 1 episode
• Adversarial Examples — 1× in this paper · explained in 2 episodes
• Benchmarking — 1× in this paper · explained in 3 episodes
• Downstream Tasks — 1× in this paper · explained in 2 episodes
• Edit Distance Similarity — 1× in this paper
• Failure Modes — 1× in this paper · explained in 3 episodes
• Foundation Model — 1× in this paper · explained in 3 episodes
• Foundation Models — 1× in this paper · explained in 7 episodes
• Grounding — 1× in this paper · explained in 3 episodes
• Instruction Tuning — 1× in this paper · explained in 2 episodes
• Tokenization — 1× in this paper · explained in 2 episodes
• Transfer Learning — 1× in this paper · explained in 10 episodes
• WER to Character Error Rate — 1× in this paper

Sources

• GPT-4 Technical Report
• Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond
• Qwen3-VL Technical Report
• Synthetic and Natural Noise Both Break Neural Machine Translation
• Nougat: Neural Optical Understanding for Academic Documents
• InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks
• PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing
• PP-OCR: A Practical Ultra Lightweight OCR System
• OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning
• Gemini: A Family of Highly Capable Multimodal Models
• Gemma 3 Technical Report
• OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotations
• olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models
• Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs
• DocVQA: A Dataset for VQA on Document Images
• On Faithfulness and Factuality in Abstractive Summarization
• Survey of Vulnerabilities in Large Language Models Revealed by Adversarial Attacks
• OpenAI GPT-5 System Card
• Qwen3.5-Omni Technical Report
• Vision Language Models are Biased

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
• VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection