The code is tested on a V100 GPU with Tensorflow 2.0.0 and Python 3.7.6.
pip install -r requirements.txt
RNA experiments on sCNN kernels use the half Normal Glorot initialization.
To train sCNN model with hyperparameter match version on RNA problem with strand16 dataset
python3 machine_HM.py --dataset strand16s --weight 5 --lr 0.002
This project is built on top of CNN for RNA problem.