prolongvid_stage1_7B

Model Summary

The ProLongVid-v1 models are 7B parameter models trained on ProLongVid_data, based on our extended Qwen2.5 language model with a context window of 256K tokens.

This prolongvid_stage1_7B model is trained on stage-1 short-video data of ProLongVid_data, based on prolongvid_image_sft_7B.

We suggest testing this model with up to 32 frames.

• Paper: For more details, please check our paper
• Repository: ruiwang2021/ProLongVid
• Languages: English, Chinese

Citation

@inproceedings{wang2025prolongvid,
  title={ProLongVid: A Simple but Strong Baseline for Long-context Video Instruction Tuning},
  author={Wang, Rui and Li, Bohao and Dai, Xiyang and Yang, Jianwei and Chen, Yi-Ling and Xing, Zhen and Yang, Yifan and Chen, Dongdong and Qiu, Xipeng and Wu, Zuxuan and others},
  booktitle={EMNLP},
  year={2025}
}