This repository contains the code, datasets, and pre-trained models as detailed in Reinforcement Learning using a Molecular Fragment Based Approach for Reaction Discovery
Requirements: All these were executed using Google Colab Pro.
Code: We have provided all the python (.ipynb) files regarding reinforcement learning (RL). 
      	Step 1: Download folder 'code' and place it in your home directory of the google drive. This 'code' folder contains all the pre-trained weights and bias as well as all python (.py) files required to execute the model.
      	Step 2: If you want to train the model from scratch, use predictor_training.ipynb for TL-based regressor, and generator_training.ipynb for generator. 
      	Step 3: For optimization through RL, use rl_Reaction_a.ipynb/ rl_Reaction_b.ipynb/ rl_Reaction_c.ipynb for the three different reactions as reported in this study.

The 'code' is given in following link,
https://drive.google.com/drive/folders/1vROYIOVLCsIErBq-I9OhdevdiyUiHkIB?usp=sharing