vapr_qwen2vl_7B

VaPR-Qwen2VL-7B

This model is an artifact of the work VaPR – Vision-language Preference alignment for Reasoning (accepted at COLM 2025). 📂 Project Website

Base Model

• Base Model: Qwen2VL 7B Instruct

Training Data

This model has been fine-tuned on the VaPR-30k subset dataset available at: VaPR-30k

Citation

If you use this model, please cite the following paper:

inproceedings{
wadhawan2025vapr,
title={Va{PR} - Vision-language Preference alignment for Reasoning},
author={Rohan Wadhawan and Fabrice Y Harel-Canada and Zi-Yi Dou and Suhaila Shakiah and Robinson Piramuthu and Nanyun Peng},
booktitle={Second Conference on Language Modeling},
year={2025},
url={https://openreview.net/forum?id=uBAubFwymy}
}