Where Hallucinations Live: A Cross-Architecture Circuit in VQ-Tokenized Vision-Language Models

arXiv:2609.29048 · cs.CV, cs.LG · Submitted 2026-09-24 · Read on arXiv

cs.CV, cs.LG

Submitted: 2026-09-24

Updated: 2026-09-24

Project page: https://shamanthak-hegde.github.io/where-hallucinations-live

Terminology

• Ablation — 69× in this paper · explained in 1 episode
• Hallucination — 37× in this paper · explained in 4 episodes
• Calibration — 13× in this paper · explained in 7 episodes
• Benchmark — 10× in this paper · explained in 2 episodes
• Transformer — 7× in this paper · explained in 2 episodes
• Accuracy — 6× in this paper · explained in 1 episode
• Quantization — 6× in this paper · explained in 12 episodes
• Vision-language models — 6× in this paper · explained in 2 episodes
• Activation Patching — 5× in this paper · explained in 2 episodes
• Embeddings — 5× in this paper · explained in 2 episodes
• Cosine Similarity — 4× in this paper · explained in 3 episodes
• Steering — 4× in this paper · explained in 2 episodes
• Hyperparameters — 3× in this paper · explained in 2 episodes
• Failure Modes — 2× in this paper · explained in 3 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Foundation Model — 2× in this paper · explained in 3 episodes
• Interpretability — 2× in this paper · explained in 15 episodes
• Robustness — 2× in this paper · explained in 5 episodes
• Tokenization — 2× in this paper · explained in 2 episodes
• Alignment — 1× in this paper · explained in 1 episode
• Foundation Models — 1× in this paper · explained in 7 episodes
• In-Context Learning — 1× in this paper · explained in 4 episodes
• Latent Space — 1× in this paper · explained in 7 episodes
• Representation Learning — 1× in this paper · explained in 2 episodes
• Vision-Language Models (VLMs) — 1× in this paper · explained in 6 episodes

Sources

• Qwen2.5-VL Technical Report
• Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling
• ANOLE: An Open, Autoregressive, Native Large Multimodal Models for Interleaved Image-Text Generation
• Taming Transformers for High-Resolution Image Synthesis
• Lumina-mGPT: Illuminate Flexible Photorealistic Text-to-Image Generation with Multimodal Generative Pretraining
• Making LLaMA SEE and Draw with SEED Tokenizer
• Reducing Hallucinations in Vision-Language Models via Latent Space Steering
• Investigating VLM Hallucination from a Cognitive Psychology Perspective: A First Step Toward Interpretation with Intriguing Observations
• UniTok: A Unified Tokenizer for Visual Generation and Understanding
• Locating and Editing Factual Associations in GPT
• Show-o: One Single Transformer to Unify Multimodal Understanding and Generation
• Chameleon: Mixed-Modal Early-Fusion Foundation Models
• Gemma: Open Models Based on Gemini Research and Technology
• AMBER: An LLM-free Multi-dimensional Benchmark for MLLMs Hallucination Evaluation
• Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
• Emu3: Next-Token Prediction is All You Need
• VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
• HaploOmni: Unified Single Transformer for Multimodal Video Understanding and Generation

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models