Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2605.14906 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913129425797120 |
|---|---|
| author | Ren, Xiyu Wang, Zhaowei Du, Yiming Xie, Zhongwei Liu, Chi Yang, Xinlin Feng, Haoyue Pan, Wenjun Zheng, Tianshi Xu, Baixuan Li, Zhengnan Song, Yangqiu Wong, Ginny See, Simon |
| author_facet | Ren, Xiyu Wang, Zhaowei Du, Yiming Xie, Zhongwei Liu, Chi Yang, Xinlin Feng, Haoyue Pan, Wenjun Zheng, Tianshi Xu, Baixuan Li, Zhengnan Song, Yangqiu Wong, Ginny See, Simon |
| contents | Memory is essential for large vision-language models (LVLMs) to handle long, multimodal interactions, with two method directions providing this capability: long-context LVLMs and memory-augmented agents. However, no existing benchmark conducts a systematic comparison of the two on questions that genuinely require multimodal evidence. To close this gap, we introduce MEMLENS, a comprehensive benchmark for memory in multimodal multi-session conversations, comprising 789 questions across five memory abilities (information extraction, multi-session reasoning, temporal reasoning, knowledge update, and answer refusal) at four standard context lengths (32K-256K tokens) under a cross-modal token-counting scheme. An image-ablation study confirms that solving MEMLENS requires visual evidence: removing evidence images drops two frontier LVLMs below 2% accuracy on the 80.4% of questions whose evidence includes images. Evaluating 27 LVLMs and 7 memory-augmented agents, we find that long-context LVLMs achieve high short-context accuracy through direct visual grounding but degrade as conversations grow, whereas memory agents are length-stable but lose visual fidelity under storage-time compression. Multi-session reasoning caps most systems below 30%, and neither approach alone solves the task. These results motivate hybrid architectures that combine long-context attention with structured multimodal retrieval. Our code is available at https://github.com/xrenaf/MEMLENS. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_14906 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language Models Ren, Xiyu Wang, Zhaowei Du, Yiming Xie, Zhongwei Liu, Chi Yang, Xinlin Feng, Haoyue Pan, Wenjun Zheng, Tianshi Xu, Baixuan Li, Zhengnan Song, Yangqiu Wong, Ginny See, Simon Computer Vision and Pattern Recognition Memory is essential for large vision-language models (LVLMs) to handle long, multimodal interactions, with two method directions providing this capability: long-context LVLMs and memory-augmented agents. However, no existing benchmark conducts a systematic comparison of the two on questions that genuinely require multimodal evidence. To close this gap, we introduce MEMLENS, a comprehensive benchmark for memory in multimodal multi-session conversations, comprising 789 questions across five memory abilities (information extraction, multi-session reasoning, temporal reasoning, knowledge update, and answer refusal) at four standard context lengths (32K-256K tokens) under a cross-modal token-counting scheme. An image-ablation study confirms that solving MEMLENS requires visual evidence: removing evidence images drops two frontier LVLMs below 2% accuracy on the 80.4% of questions whose evidence includes images. Evaluating 27 LVLMs and 7 memory-augmented agents, we find that long-context LVLMs achieve high short-context accuracy through direct visual grounding but degrade as conversations grow, whereas memory agents are length-stable but lose visual fidelity under storage-time compression. Multi-session reasoning caps most systems below 30%, and neither approach alone solves the task. These results motivate hybrid architectures that combine long-context attention with structured multimodal retrieval. Our code is available at https://github.com/xrenaf/MEMLENS. |
| title | MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.14906 |