DiM-TS: Bridge the Gap between Selective State Space Models and Time Series for Generative Modeling
November 28, 2025 ยท View on GitHub
The repo is the official implementation for the paper: https://arxiv.org/pdf/2511.18312
DiM-TS Architecture
Running the Code
The following instructions explain how to run the code.
Environment & Libraries
The full libraries list is provided as requirements.txt. Please create a virtual environment and run
pip install -r requirements.txt
Install dependent packages
pip install --upgrade pip
pip install -r requirements.txt
cd kernels/selective_scan && pip install .
cd kernels/dwconv2d && python3 setup.py install --user
Training
For training, you can reproduce the experimental results by runing
python main.py --name {name} --config_file {config.yaml} --gpu 0 --train
Unconstrained Sampling
Please use the saved model for sampling by running
python main.py --name {name} --config_file {config.yaml} --gpu 0 --sample 0 --milestone {checkpoint_number}
Channel Permutation Scanning
channel_permutation.ipynb provides an example code of reproducing Permutation Scanning Algorithm. You can modify the content according to your dataset requirements.
Citation
If you find this repo useful, please cite our paper!
@article{yao2025dim,
title={DiM-TS: Bridge the Gap between Selective State Space Models and Time Series for Generative Modeling},
author={Yao, Zihao and Zuo, Jiankai and Zhang, Yaying},
journal={arXiv preprint arXiv:2511.18312},
year={2025}
}
Code
Thanks for the open sources papers listed below which DiM-TS is build on.
https://github.com/wmd3i/PaD-TS