A Diagnosis-based Framework for Efficient Deep Learning Hyperparameter Optimization.
Based on Python 3.10
git clone [email protected]:Dmagine/BTTackler.git
cd BTTackler
pip install -r requirements.txt
pip install -U .
cd cases
bash example_run_case.sh
or just,
cd cases
# random
python run_case.py cifar10cnn random
# bttackler-random
python run_case.py cifar10cnn random_bttackler
# all for cifar10cnn
python run_case.py cifar10cnn all
- Higher code readability and configurable quality indicators.
- Code formatting and complete experiment cases.
- Cases for calibrating quality indicators.
If you find BTTackler useful, please cite our paper.
@inproceedings{pei2024bttackler,
title={BTTackler: A Diagnosis-based Framework for Efficient Deep Learning Hyperparameter Optimization},
author={Zhongyi Pei and Zhiyao Cen and Yipeng Huang and Chen Wang and Lin Liu and Philip Yu and Mingsheng Long and Jianmin Wang},
booktitle={In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
year={2024},
}
If you have any questions or suggestions, feel free to contact:
- Zhongyi Pei ([email protected])
- Zhiyao Cen ([email protected])
Or describe it in Issues.