This source code is exploited to support the work in "Generalization Bounds for Estimating Causal Effects of Continuous Treatments", submitted to NeurIPS 2022. We provide a PyTorch implementation of the ADMIT model for estimating the causal effects of continuous treatments, i.e., the average dose-response function (ADRF).

Please create a python project in your local workstation, and import these files contained in the project.

For example, this command trains an ADMIT model on the Simulation dataset with GPU 0.

python main.py --data sim  --learning_rate 0.0002 --batch_size 500 --log

If you find our work useful for your research, please consider citing the following papers :)

@inproceedings{wang2022generalization,
  title={Generalization Bounds for Estimating Causal Effects of Continuous Treatments},
  author={Xin Wang and Shengfei Lyu and Xingyu Wu and Tianhao Wu and Huanhuan Chen},
  booktitle={NeurIPS},
  year={2022}
}