Bridging the Gap: Teacher-Assisted Wasserstein Knowledge Distillation for Efficient Multi-Modal Recommendation

Introduction

This is the Pytorch implementation for our WWW 2025 paper:

[WWW ’25] Ziyi Zhuang, Hanwen Du, Hui Han, Youhua Li, Junchen Fu, Joemon M. Jose, and Yongxin Ni. Bridging the Gap: Teacher‑Assisted Wasserstein Knowledge Distillation for Efficient Multi‑Modal Recommendation. In Proceedings of the ACM Web Conference 2025 (WWW ’25), April 28–May 2, 2025, Sydney, NSW, Australia. ACM, New York, NY, USA. 12 pages. DOI: 10.1145/3696410.3714852

Getting Started

To train the model, you can easily achieve results by running the following command:

python TARec/scripts/tiktok/train_tiktok_module_ab.sh

Citation

If you find Guider useful in your research, please consider citing our paper.

@inproceedings{zhuang2025bridging,
  title={Bridging the Gap: Teacher-Assisted Wasserstein Knowledge Distillation for Efficient Multi-Modal Recommendation},
  author={Zhuang, Ziyi and Du, Hanwen and Han, Hui and Li, Youhua and Fu, Junchen and Jose, Joemon M and Ni, Yongxin},
  booktitle={THE WEB CONFERENCE 2025},
  year={2025}
}

This code is made available solely for academic research purposes.