Research map / Vision-language and multimodal / Multimodal LLMs
Why LVLMs Are More Prone to Hallucinations in Longer Responses: The Role of Context
Ge Zheng, Jiaye Qian, Jiajin Tang, Sibei Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add External Expansion detection (ablation) | HalTrapper (LLaVA v1.5 7B, nucleus) · CHAIR | CHAIR_S: 58.6 -> 51.0 | Improved |
| add Internal Grounding detection (ablation) | HalTrapper (LLaVA v1.5 7B, nucleus) · CHAIR | CHAIR_S: 58.6 -> 50.4 | Improved |
| combine IG and EE (ablation) | HalTrapper (LLaVA v1.5 7B, nucleus) · CHAIR | CHAIR_S: 51.0 -> 48.6 | Improved |
| add IG on top of EE (ablation) | HalTrapper (LLaVA v1.5 7B, nucleus) · CHAIR | CHAIR_I: 14.4 -> 14.5 | No clear effect |
| apply HalTrapper | LLaVA v1.5 7B greedy · CHAIR | CHAIR_S: 52.2 -> 41.6 | Improved |
| apply HalTrapper | LLaVA v1.5 7B nucleus · CHAIR | CHAIR_S: 58.6 -> 48.6 | Improved |
| apply HalTrapper | LLaVA v1.5 7B beam search · CHAIR | CHAIR_S: 55.6 -> 45.2 | Improved |
| apply HalTrapper | LLaVA v1.5 7B greedy · CHAIR | Recall: 80.3 -> 80.1 | No clear effect |
| apply HalTrapper | Qwen2 VL · AMBER generative | CHAIR: 6.6 -> 5.6 | Improved |
| apply HalTrapper | Qwen2 VL · AMBER generative | Cover: 71.8 -> 70.9 | Hurt |
| apply ICD decoding | Qwen2 VL · AMBER generative | Hal: 50.3 -> 74.9 | Hurt |
| apply VCD decoding | Qwen2 VL · AMBER generative | CHAIR: 6.6 -> 7.3 | Hurt |
Closest papers by meaning
- Toward a Stable, Fair, and Comprehensive Evaluation of Object Hallucination in Large Vision-Language Models (NeurIPS 2024 · similarity 0.92)
- Combating Multimodal LLM Hallucination via Bottom-Up Holistic Reasoning (AAAI 2025 · similarity 0.92)
- Prefill-Time Intervention for Mitigating Hallucination in Large Vision-Language Models (CVPR 2026 · similarity 0.92)
- Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens (CVPR 2025 · similarity 0.91)
- The Mirage of Performance Gains: Why Contrastive Decoding Fails to Mitigate Object Hallucinations in MLLMs? (NeurIPS 2025 · similarity 0.91)
- Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering (ICLR 2025 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
