Research map / Vision-language and multimodal / Multimodal LLMs
From Macro to Micro: Benchmarking Microscopic Spatial Intelligence on Molecules via Vision-Language Models
Zongzhao Li, Xiangzhe Kong, Jiahui Su, Zongyang Ma, Mingze Li, Songyou Li, Yuelin Zhang, Yu Rong, Tingyang Xu, Deli Zhao, Wenbing Huang
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 |
|---|---|---|---|
| fine-tune Qwen2.5-VL-7B on MiSI-Bench training split | Qwen2.5-VL-7B · MiSI-Bench(tiny) average | score: 34.37 -> 62.96 | Improved |
| fine-tune on MiSI-Bench | Qwen2.5-VL-7B · Residue-ligand Interaction (Pos) and Pocket-ligand Interaction | score: Pocket-ligand 10.72 vs human 82.78 | Mixed |
| replace molecular images with random images (text-only) (ablation) | Qwen2.5VL-7B-SFT · MiSI-Bench T1 Translation | score: 99.8 -> 33.5 | Hurt |
| replace molecular images with random images (text-only) (ablation) | Qwen2.5VL-7B-SFT · MiSI-Bench T9 Pocket-ligand | score: 10.7 -> 0.0 | Hurt |
| paraphrase instruction templates with GPT-5.2 (ablation) | Qwen2.5VL-7B-SFT · MiSI-Bench T1 Translation | score: 99.8 -> 99.7 | No clear effect |
| paraphrase instruction templates (ablation) | Qwen2.5VL-7B-SFT · MiSI-Bench T3 Zooming | score: 27.1 -> 29.4 | No clear effect |
| reduce input from three orthogonal views to a single view (ablation) | Qwen2.5VL-7B-SFT · Rotation | score: 99.71 -> 97.51 | No clear effect |
| increase views from 6 to 27 (ablation) | Qwen2.5VL-7B-SFT · Pocket-ligand Interaction | score: 10.72 -> 23.77 | Improved |
| increase number of hydrogen bonds in sample (ablation) | Qwen2.5VL-7B-SFT · Residue-ligand Interaction | exact match accuracy: 89.52 -> 20.51 | Hurt |
Closest papers by meaning
- Micro-Bench: A Microscopy Benchmark for Vision-Language Understanding (NeurIPS 2024 · similarity 0.88)
- TinyChemVL: Advancing Chemical Vision-Language Models via Efficient Visual Token Reduction and Complex Reaction Tasks (AAAI 2026 · similarity 0.87)
- MicroVQA: A Multimodal Reasoning Benchmark for Microscopy-Based Scientific Research (CVPR 2025 · similarity 0.86)
- nach0-pc: Multi-task Language Model with Molecular Point Cloud Encoder (AAAI 2025 · similarity 0.85)
- LTD-Bench: Evaluating Large Language Models by Letting Them Draw (NeurIPS 2025 · similarity 0.85)
- Improving Large Molecular Language Model via Relation-aware Multimodal Collaboration (AAAI 2026 · similarity 0.85)
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.
