Skip to main content

Table 1 Property prediction model performance

From: Molecular optimization by capturing chemist’s intuition using deep neural networks

  LogD Solubility HLM CLint
Train size 170,337 184,883 144,300
Train RMSE 0.304 0.485 0.264
Train NRMSE 0.041 0.079 0.083
Train \(R^2\) 0.935 0.774 0.749
Test size 18,927 20,543 16,034
Test RMSE 0.395 0.602 0.350
Test NRMSE 0.054 0.104 0.113
Test \(R^2\) 0.892 0.658 0.557