Research map / Vision-language and multimodal / Visual question answering
AVAM: a Universal Training-free Adaptive Visual Anchoring Embedded into Multimodal Large Language Model for Multi-image Question Answering
Kang Zeng, Guojin Zhong, Jintao Cheng, Jin Yuan, Zhiyong Li
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 Implicit Retrieval token pruning (R=0.1) (ablation) | LLaVA-v1.5-7B · MuirBench | accuracy (avg): 24.2 -> 26.0 | Improved |
| raise IR retention ratio R from 0.1 to 0.5 (ablation) | LLaVA-v1.5-7B · MuirBench | accuracy (avg): 26.0 -> 27.0 | Improved |
| add contrastive denoising-style CD (lambda=5) on top of IR (ablation) | LLaVA-v1.5-7B · MuirBench | accuracy (avg): 27.0 -> 27.6 | Improved |
| set CD lambda to 1 instead of 5 (ablation) | LLaVA-v1.5-7B · MuirBench | accuracy (avg): 27.6 -> 25.2 | Hurt |
| plug AVAM into LLaVA-v1.5-7B | LLaVA-v1.5-7B · MuirBench | accuracy (avg): 24.2 -> 27.6 | Improved |
| plug AVAM into InternVL2-8B | InternVL2-8B · MuirBench | accuracy (avg): 36.6 -> 39.5 | Improved |
| plug AVAM into Mantis-8B | Mantis-8B · MuirBench | accuracy (avg): 34.0 -> 34.5 | Improved |
| plug AVAM into Qwen-VL-7B | Qwen-VL-7B · MuirBench | accuracy (avg): 13.6 -> 14.0 | Improved |
| plug AVAM into LLaVA-v1.5-7B | LLaVA-v1.5-7B · MIBench MII | accuracy (avg): 16.2 -> 31.7 | Improved |
| plug AVAM into DeepSeek-VL-7B | DeepSeek-VL-7B · MIBench MII | accuracy (avg): 33.4 -> 52.0 | Improved |
| plug AVAM into LLaVA-OV-Qwen2-7B | LLaVA-OV-Qwen2-7B · MIBench MII | accuracy (avg): 67.2 -> 68.2 | Improved |
| plug AVAM into InternVL2-8B | InternVL2-8B · Mantis-Eval | accuracy: 55.4 -> 59.7 | Improved |
Closest papers by meaning
- Efficient Large Multi-modal Models via Visual Context Compression (NeurIPS 2024 · similarity 0.91)
- Rethinking Token Reduction for Large Vision-Language Models (CVPR 2026 · similarity 0.91)
- FOCUS: Internal MLLM Representations for Efficient Fine-Grained Visual Question Answering (NeurIPS 2025 · similarity 0.90)
- BLIVA: A Simple Multimodal LLM for Better Handling of Text-Rich Visual Questions (AAAI 2024 · similarity 0.90)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
- CL-MoE: Enhancing Multimodal Large Language Model with Dual Momentum Mixture-of-Experts for Continual Visual Question Answering (CVPR 2025 · similarity 0.90)
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.
