---
license: mit
---

# DDBC Checkpoints

This repository provides pretrained checkpoints for **DDBC (Discrete Diffusion for Bundle Construction)**.

The checkpoints correspond to models trained on the following datasets:

- **pog_dense**
- **spotify30**
- **spotify60**
- **spotify90**

## Paper

For details about the method, please refer to the paper:

**Discrete Diffusion for Bundle Construction**  
ICLR 2026

OpenReview:  
https://openreview.net/forum?id=dKyhgfe50H

## Code

The official implementation is available on GitHub:

https://github.com/241416/DDBC

## Checkpoints

The following pretrained checkpoints are included in this repository:

| Dataset | Checkpoint |
|-------|-------|
| pog_dense | `pog_dense.ckpt` |
| spotify30 | `spotify30.ckpt` |
| spotify60 | `spotify60.ckpt` |
| spotify90 | `spotify90.ckpt` |

These checkpoints can be used for evaluation or further fine-tuning with the provided codebase.

## Citation

If you use the code or checkpoints in your research, please cite:

```bibtex
@inproceedings{tu2026discrete,
  title={Discrete Diffusion for Bundle Construction},
  author={Tu, Teng and Li, Ai and Ma, Yunshan and Xu, Shuo and Liu, Xiaohao and Ma, Haokai and Pang, Liang and Chua, Tat-Seng},
  booktitle={The Fourteenth International Conference on Learning Representations},
  year={2026},
  url={https://openreview.net/forum?id=dKyhgfe50H}
}