Request for Video-R1 3B model training data and weights

[chengY334]

Hi authors,
I am a graduate student. I am currently working on reproducing and extending the Video-R1 framework, and I am deeply impressed by your pioneering work on R1 for video reasoning.
Due to limited computational resources, my research is focused on the 3B model size. I noticed that the released model weights are for the 7B version, so I would like to kindly ask if you could share the 3B model version and its corresponding training data. This would be extremely helpful for my academic research.
Thank you very much for your time and valuable contribution to the community.
Best regards,

[tulerfeng]

Hi, I have not trained 3B scale models.

You may also use our data to tune the Qwen2.5-VL-3B directly, just by revising the model name.