Research map / Vision-language and multimodal / Multimodal LLMs
INTER: Mitigating Hallucination in Large Vision-Language Models by Interaction Guidance Sampling
Xin Dong, Shichao Dong, Jin Wang, Jing Huang, Li Zhou, Zenghui Sun, Lihua Jing, Jinsong Lan, Xiaoyong Zhu, Bo Zheng
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 |
|---|---|---|---|
| apply INTER on top of nucleus sampling | InstructBLIP · MME total | total score: 1251.8 -> 1595.5 | Improved |
| apply INTER on top of nucleus sampling | LLaVA-v1.5 7B · POPE COCO | avg F1: 79.7 -> 85.7 | Improved |
| apply INTER on top of nucleus sampling | LLaVA-v1.5 7B · CHAIR 64 tokens | CHAIR_S: 24.4 -> 19.8 | Improved |
| apply INTER on top of OPERA | InstructBLIP · CHAIR 64 tokens | CHAIR_I: 6.8 -> 7.9 | Hurt |
| apply INTER on top of OPERA | Qwen-VL · POPE AOKVQA | avg F1: 85.1 -> 84.4 | Hurt |
| apply INTER on top of VCD | InstructBLIP · POPE AOKVQA | avg F1: 81.0 -> 80.8 | Hurt |
| apply INTER on top of beam search | LLaVA-v1.5 7B · MM-Bench | overall accuracy: 65.1 -> 65.1 | No clear effect |
| apply INTER on top of nucleus sampling | LLaVA-v1.5 7B · MM-Bench | overall accuracy: 57.3 -> 62.6 | Improved |
| apply INTER on top of greedy search | mPLUG-owl2 · MM-Bench | overall accuracy: 63.5 -> 62.3 | Hurt |
| add Interactive Guided Locator to Interaction Probability Modifier (ablation) | INTER · MME total (LLaVA-v1.5 7B) | total score: 1648.6 -> 1744.0 | Improved |
| apply modifier at all steps (k = 0) instead of only key tokens (k = 1) (ablation) | INTER · CHAIR long captions | CHAIR | Hurt |
| vary threshold k within 0.5 to 1.5 (ablation) | INTER · various LVLMs | performance | No clear effect |
Closest papers by meaning
- Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens (CVPR 2025 · similarity 0.94)
- Combating Multimodal LLM Hallucination via Bottom-Up Holistic Reasoning (AAAI 2025 · similarity 0.94)
- Multi-Object Hallucination in Vision Language Models (NeurIPS 2024 · similarity 0.93)
- Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering (ICLR 2025 · similarity 0.93)
- AVHBench: A Cross-Modal Hallucination Benchmark for Audio-Visual Large Language Models (ICLR 2025 · similarity 0.93)
- Prefill-Time Intervention for Mitigating Hallucination in Large Vision-Language Models (CVPR 2026 · similarity 0.93)
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.
