Pelican-VL 1.0: A Foundation Brain Model for Embodied Intelligence
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909901641482240 |
|---|---|
| author | Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Hanzhe Niu, Zhenwei Liu, Zhaoyang Liu, Shuang Zhao, Yue Qi, Junbo Zhang, Qinfan Li, Dengjie Wang, Yidong Luo, Jiachen Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu |
| author_facet | Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Hanzhe Niu, Zhenwei Liu, Zhaoyang Liu, Shuang Zhao, Yue Qi, Junbo Zhang, Qinfan Li, Dengjie Wang, Yidong Luo, Jiachen Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu |
| contents | This report presents Pelican-VL 1.0, a new family of open-source embodied brain models with parameter scales ranging from 7 billion to 72 billion. Our explicit mission is clearly stated as: To embed powerful intelligence into various embodiments. Pelican-VL 1.0 is currently the largest-scale open-source embodied multimodal brain model. Its core advantage lies in the in-depth integration of data power and intelligent adaptive learning mechanisms. Specifically, metaloop distilled a high-quality dataset from a raw dataset containing 4+ billion tokens. Pelican-VL 1.0 is trained on a large-scale cluster of 1000+ A800 GPUs, consuming over 50k+ A800 GPU-hours per checkpoint. This translates to a 20.3% performance uplift from its base model and outperforms 100B-level open-source counterparts by 10.6%, placing it on par with leading proprietary systems on well-known embodied benchmarks. We establish a novel framework, DPPO (Deliberate Practice Policy Optimization), inspired by human metacognition to train Pelican-VL 1.0. We operationalize this as a metaloop that teaches the AI to practice deliberately, which is a RL-Refine-Diagnose-SFT loop. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_00108 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Pelican-VL 1.0: A Foundation Brain Model for Embodied Intelligence Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Hanzhe Niu, Zhenwei Liu, Zhaoyang Liu, Shuang Zhao, Yue Qi, Junbo Zhang, Qinfan Li, Dengjie Wang, Yidong Luo, Jiachen Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu Machine Learning Artificial Intelligence Robotics This report presents Pelican-VL 1.0, a new family of open-source embodied brain models with parameter scales ranging from 7 billion to 72 billion. Our explicit mission is clearly stated as: To embed powerful intelligence into various embodiments. Pelican-VL 1.0 is currently the largest-scale open-source embodied multimodal brain model. Its core advantage lies in the in-depth integration of data power and intelligent adaptive learning mechanisms. Specifically, metaloop distilled a high-quality dataset from a raw dataset containing 4+ billion tokens. Pelican-VL 1.0 is trained on a large-scale cluster of 1000+ A800 GPUs, consuming over 50k+ A800 GPU-hours per checkpoint. This translates to a 20.3% performance uplift from its base model and outperforms 100B-level open-source counterparts by 10.6%, placing it on par with leading proprietary systems on well-known embodied benchmarks. We establish a novel framework, DPPO (Deliberate Practice Policy Optimization), inspired by human metacognition to train Pelican-VL 1.0. We operationalize this as a metaloop that teaches the AI to practice deliberately, which is a RL-Refine-Diagnose-SFT loop. |
| title | Pelican-VL 1.0: A Foundation Brain Model for Embodied Intelligence |
| topic | Machine Learning Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2511.00108 |