Skip to content

Latest commit

 

History

History
13 lines (8 loc) · 696 Bytes

File metadata and controls

13 lines (8 loc) · 696 Bytes
  1. Dateset: NNrate/data/

All the datasets used in this work are given in the file named "dateset".

  1. Neural network: NNrate/train/

The codes for training is named "nn_rate.py". The training date set must be formated as the following five numbers in each line, X1, X2, X3,the index of carbon atom of the broken C-H bond and temperature. An example of input files is supplied as "testset" and "trainset". The optimized parameters are given in the file "saved_weights.pth".

  1. Prediction: NNrate/predict/

Running nn_predict.py with the provided neural network parameters and descriptors of the target reaction, the predicted rate constant from each model (or scheme) will be supplied.