Extended Data Fig. 9: Training iterations for LiCl assembly.
From: Machine-guided path sampling to discover mechanisms of molecular self-organization
Training iterations for Li+ Cl− assembly. The blue line shows the learning rate calculated from the efficiency factor at every step, orange crosses show when training actually occurred. The inset shows the training loss per shooting point for every training. Only the first 26000 Monte Carlo steps are shown. The model uses the inputs ‘SF longranged II’ and is of architecture ‘ResNet I’ (see Supplementary Tables 5 and 6).
