Innovator-VL: A Multimodal Large Language Model for Scientific Discovery
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912852074299392 |
|---|---|
| author | Wen, Zichen Yang, Boxue Chen, Shuang Zhang, Yaojie Han, Yuhang Ke, Junlong Wang, Cong Fu, Yicheng Zhao, Jiawang Yao, Jiangchao Fang, Xi Wang, Zhen Cai, Henxing Yao, Lin Gao, Zhifeng Hong, Yanhui Yuan, Nang Li, Yixuan Zhao, Guojiang Tao, Haoyi Wang, Nan Lyu, Han Ke, Guolin Liao, Ning Wang, Xiaoxing Chen, Kai Li, Zhiyu Xiong, Feiyu Hu, Sihan Chen, Kun Wang, Yanfeng E, Weinan Zhang, Linfeng Zhang, Linfeng |
| author_facet | Wen, Zichen Yang, Boxue Chen, Shuang Zhang, Yaojie Han, Yuhang Ke, Junlong Wang, Cong Fu, Yicheng Zhao, Jiawang Yao, Jiangchao Fang, Xi Wang, Zhen Cai, Henxing Yao, Lin Gao, Zhifeng Hong, Yanhui Yuan, Nang Li, Yixuan Zhao, Guojiang Tao, Haoyi Wang, Nan Lyu, Han Ke, Guolin Liao, Ning Wang, Xiaoxing Chen, Kai Li, Zhiyu Xiong, Feiyu Hu, Sihan Chen, Kun Wang, Yanfeng E, Weinan Zhang, Linfeng Zhang, Linfeng |
| contents | We present Innovator-VL, a scientific multimodal large language model designed to advance understanding and reasoning across diverse scientific domains while maintaining excellent performance on general vision tasks. Contrary to the trend of relying on massive domain-specific pretraining and opaque pipelines, our work demonstrates that principled training design and transparent methodology can yield strong scientific intelligence with substantially reduced data requirements. (i) First, we provide a fully transparent, end-to-end reproducible training pipeline, covering data collection, cleaning, preprocessing, supervised fine-tuning, reinforcement learning, and evaluation, along with detailed optimization recipes. This facilitates systematic extension by the community. (ii) Second, Innovator-VL exhibits remarkable data efficiency, achieving competitive performance on various scientific tasks using fewer than five million curated samples without large-scale pretraining. These results highlight that effective reasoning can be achieved through principled data selection rather than indiscriminate scaling. (iii) Third, Innovator-VL demonstrates strong generalization, achieving competitive performance on general vision, multimodal reasoning, and scientific benchmarks. This indicates that scientific alignment can be integrated into a unified model without compromising general-purpose capabilities. Our practices suggest that efficient, reproducible, and high-performing scientific multimodal models can be built even without large-scale data, providing a practical foundation for future research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_19325 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Innovator-VL: A Multimodal Large Language Model for Scientific Discovery Wen, Zichen Yang, Boxue Chen, Shuang Zhang, Yaojie Han, Yuhang Ke, Junlong Wang, Cong Fu, Yicheng Zhao, Jiawang Yao, Jiangchao Fang, Xi Wang, Zhen Cai, Henxing Yao, Lin Gao, Zhifeng Hong, Yanhui Yuan, Nang Li, Yixuan Zhao, Guojiang Tao, Haoyi Wang, Nan Lyu, Han Ke, Guolin Liao, Ning Wang, Xiaoxing Chen, Kai Li, Zhiyu Xiong, Feiyu Hu, Sihan Chen, Kun Wang, Yanfeng E, Weinan Zhang, Linfeng Zhang, Linfeng Computer Vision and Pattern Recognition Artificial Intelligence We present Innovator-VL, a scientific multimodal large language model designed to advance understanding and reasoning across diverse scientific domains while maintaining excellent performance on general vision tasks. Contrary to the trend of relying on massive domain-specific pretraining and opaque pipelines, our work demonstrates that principled training design and transparent methodology can yield strong scientific intelligence with substantially reduced data requirements. (i) First, we provide a fully transparent, end-to-end reproducible training pipeline, covering data collection, cleaning, preprocessing, supervised fine-tuning, reinforcement learning, and evaluation, along with detailed optimization recipes. This facilitates systematic extension by the community. (ii) Second, Innovator-VL exhibits remarkable data efficiency, achieving competitive performance on various scientific tasks using fewer than five million curated samples without large-scale pretraining. These results highlight that effective reasoning can be achieved through principled data selection rather than indiscriminate scaling. (iii) Third, Innovator-VL demonstrates strong generalization, achieving competitive performance on general vision, multimodal reasoning, and scientific benchmarks. This indicates that scientific alignment can be integrated into a unified model without compromising general-purpose capabilities. Our practices suggest that efficient, reproducible, and high-performing scientific multimodal models can be built even without large-scale data, providing a practical foundation for future research. |
| title | Innovator-VL: A Multimodal Large Language Model for Scientific Discovery |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2601.19325 |