Skip to main content

Table 3 Property prediction model performance on in-house data

From: Transformer-based molecular optimization beyond matched molecular pairs

  LogD Solubility Clearance
Train size 186,575 197,988 155,652
Train RMSE 0.295 0.489 0.271
Train NRMSE 0.025 0.056 0.053
Train \(R^2\) 0.942 0.775 0.760
Test size 20,731 21,999 17,295
Test RMSE 0.395 0.600 0.352
Test NRMSE 0.038 0.076 0.091
Test \(R^2\) 0.897 0.659 0.555