B. Liu, S. Gao, X. Liu, X. Cheng, L. Yang. “WiFo: Wireless Foundation Model for Channel Prediction. ” SCIENCE CHINA Information Sciences, June 2025, 68(6): 162302.
🔥🔥🔥 Last Updated on 2025.11.27 🔥🔥🔥
- [2025.11.28] We released all the pre-training datasets as open source at [Pretraining Dataset].
- Python 3.9 (Recommend to use Anaconda)
- Pytorch 2.0.0
- NVIDIA GPU + CUDA
- Python packages:
pip install -r requirements.txt
The dataset used for inference is available in [Testing Dataset]. The dataset used for pre-training is available in [Pretraining Dataset]
We have released the pre-trained weights of WiFo-Large/Base/Small/Little/Tiny for inference in [Model].
Note: For the large model variant, both the encoder and decoder Transformers use 12 attention heads, rather than the 8 heads stated in the paper.
python main.py --device_id 1 --size base --mask_strategy_random none --mask_strategy temporal --dataset D1*D2*D3*D4*D5*D6*D7*D8*D9*D10*D11*D12*D13*D14*D15*D16 --file_load_path ./weights/wifo_base --few_ratio 0.0 --t_patch_size 4 --patch_size 4 --batch_size 128 --pos_emb SinCos_3D
python main.py --device_id 1 --size base --mask_strategy_random none --mask_strategy fre --dataset D1*D2*D3*D4*D5*D6*D7*D8*D9*D10*D11*D12*D13*D14*D15*D16 --file_load_path ./weights/wifo_base --few_ratio 0.0 --t_patch_size 4 --patch_size 4 --batch_size 128 --pos_emb SinCos_3D
python main.py --device_id 1 --size base --mask_strategy_random none --mask_strategy temporal --dataset D17 --file_load_path ./weights/wifo_base --few_ratio 0.0 --t_patch_size 4 --patch_size 4 --batch_size 128 --pos_emb SinCos_3D
If you find this repo helpful, please cite our paper.
@article{liu2025wifo,
author = {Liu, Boxun and Gao, Shijian and Liu, Xuanyu and Cheng, Xiang and Yang, Liuqing},
title = {{WiFo: Wireless Foundation Model for Channel Prediction}},
journal = {Sci. China Inf. Sci.},
volume = {68},
pages = {162302},
year = {2025},
month = {May},
doi = {10.1007/s11432-025-4349-0}
}