---
license: mit
library_name: pytorch
pipeline_tag: video-text-to-text
tags:
  - video-understanding
  - multimodal
---

# HierarQ Checkpoint

Checkpoint for **HierarQ** from the paper *HierarQ: Task-Aware Hierarchical Q-Former for Enhanced Video
Understanding* (CVPR, 2025).

- **Paper:** [ArXiv](https://arxiv.org/pdf/2503.08585)
- **Code:** [GitHub link](https://github.com/sacrcv/HierarQ)

## Contents
- `hierarq_msvd_cap.pth`: This is an intermediate checkpoint of the model trained on the MSVD dataset on the captioning task

## Usage
Download the checkpoint and follow the instructions in the code repository:

    hf download shehreen3764/HierarQ_Checkpoint --local-dir ./checkpoints

## Citation
```bibtex
@inproceedings{azad2025hierarq,
  title     = {HierarQ: Task-Aware Hierarchical Q-Former for Enhanced Video Understanding},
  author    = {Azad, Shehreen and Vineet, Vibhav and Rawat, Yogesh Singh},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2025}
}
```