Skip to main content

Table 1 Comparison architecture A and B and comparing LSTM to GRU

From: GEN: highly efficient SMILES explorer using autodidactic generative examination networks

ArchitectureLayer sizeBest model epoch#Validity%Uniqueness%Training%Length match %aHAC match %b
A: LSTM–LSTM256/25612, 17, 2096.7 ± 0.499.9 ± 0.115.0 ± 0.798.2 ± 0.994.0 ± 1.8
A: GRU–GRU256/25615, 15, 1591.8 ± 0.799.9 ± 0.112.6 ± 0.898.3 ± 0.494.6 ± 1.3
B: biLSTM–biLSTM256/2566, 7, 1097.1 ± 0.499.9 ± 0.113.1 ± 0.598.2 ± 0.693.9 ± 0.8
B: biGRU–biGRU256/25611, 11, 1195.6 ± 0.699.9 ± 0.115.0 ± 0.598.3 ± 0.393.1 ± 1.4
  1. aLength match for SMILES length distributions of the training set and generated set (See “Methods”)
  2. bHAC match for the atom count distributions of the generated set and training set (See “Methods”)