From: Predicting chemical structure using reinforcement learning with a stack-augmented conditional variational autoencoder
% of valid SMILES
% of valid and unique SMILES
% of valid and unique SMILES not in training data
Stack-CVAE (%)
98.1
81.1
CVAE (%)
39.9
SSVAE (%)
98.8
16.6
Stack-RNN (%)
100
1.5
1.2
cRNN (%)
99.7
25.2