This directory contains the rmse and mae values (kJ/mol) for training a GPR on the chlorpropamide dataset as well as the names of training set/test set structures. Errors files are provided for each kernel type and cut-off.



Error Data is supplied as arrays, each entry - in order- is another 'step' of training. That is, the first entry in any array gives the error when training on 500 structures, the second gives the error after training on 1000 structures.



Training set names files are supplied as arrays of names. Filenames can be complex, but the key 'root' - e.g confD_5_43798, identifies the structure. Progressive training is set up such that the first 500 training structures are train_set_names_1, the first 100 training structures are train_set_names_1 and train_set_names_2, etc.

A text file with the correct calculated DFT energies of the structures is also provided.
