Spatial-Temporal Graph Transformer (Pretrained) - LARGEST-GLA

Spatial-Temporal Graph Transformer (Pretrained) (STGFORMER_PRETRAINED) trained on LARGEST-GLA dataset for traffic speed forecasting.

Model Description

STGFormer pretrained checkpoint for LARGEST-GLA. This checkpoint contains pretrained model weights and imputation head from masked node pretraining. Use with load_from config option.

Dataset

LARGEST-GLA: Traffic speed data from highway sensors.

Usage

from utils.stgformer import load_from_hub

# Load model from Hub
model, scaler = load_from_hub("LARGEST-GLA", hf_repo_prefix="STGFORMER_PRETRAINED")

# Get predictions
from utils.stgformer import get_predictions
predictions = get_predictions(model, scaler, test_dataset)

Training

Model was trained using the STGFORMER_PRETRAINED implementation with default hyperparameters.

Citation

If you use this model, please cite the original STGFORMER_PRETRAINED paper:

@inproceedings{lan2022stgformer,
  title={STGformer: Spatial-Temporal Graph Transformer for Traffic Forecasting},
  author={Lan, Shengnan and Ma, Yong and Huang, Weijia and Wang, Wanwei and Yang, Hui and Li, Peng},
  booktitle={IEEE Transactions on Neural Networks and Learning Systems},
  year={2022}
}

License

This model checkpoint is released under the same license as the training code.

Downloads last month

-

Downloads are not tracked for this model. How to track
Safetensors
Model size
5.29M params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support