AirFM-DDA
Pretrained weights for AirFM-DDA: Air-Interface Foundation Model in the Delay-Doppler-Angle Domain for AI-Native 6G (IEEE TNSE, accepted).
Code, evaluation data and instructions: github.com/SII-kejia/AirFM-DDA
| File | Variant | Embedding dim | Heads | Depths |
|---|---|---|---|---|
small.pt |
small | 512 | 4 | 6, 6, 6, 6 |
base.pt |
base | 640 | 5 | 6, 6, 6, 6 |
large.pt |
large | 768 | 6 | 6, 6, 6, 6 |
Each file is a PyTorch state dict. The matching model configurations are models/{small,base,large}.json in the GitHub repository.
Usage
git clone https://github.com/SII-kejia/AirFM-DDA.git
cd AirFM-DDA
pip install -e . huggingface_hub
hf download SII-kejia/AirFM-DDA-Model small.pt base.pt large.pt --local-dir models
bash scripts/evaluate.sh
See the GitHub README for evaluation, fine-tuning and downstream heads.
Citation
@article{bian2026airfmdda,
title = {{AirFM-DDA}: Air-Interface Foundation Model in the Delay-Doppler-Angle Domain for {AI}-Native {6G}},
author = {Bian, Kejia and Tao, Meixia and Mo, Jianhua and Chen, Zhiyong and Chen, Leyan},
journal = {IEEE Transactions on Network Science and Engineering},
year = {2026},
note = {Accepted}
}
License
MIT
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support