This directory contains the key ML data - e.g errors - for training GPR models on small structure sets of chlorpropamide, target 31, and target 32.

Important note: These .txt. files ar einfact pickle files and must be loaded using the pickle package in python before reading

The dataset for each system contains (for each kernel type and cut off tested):

sub_rmses/maes.txt files - these contain dictionaries giving training set sizes as keys and lists of errors (across all cross validation iterations) as values
av_maes.txt files - these contain dictionaries giving training set sizes as keys and average errors (averaged across validation folds) as values


filenaming can be complex, but the key roots of the filenames will identify the system, the kernel type, and SOAP cut-offs used.



