Importing a pre-computed training set

A training set is a set of text files containing rating features and manual user ratings. Training sets are the basis of the rating scheme employed in PyJibe (and nanite) [MAM+19]. By default, PyJibe comes with the zef18 training set [MMG18]. For general information about how manual rating and generation of the training set works, please refer to the nanite rating workflow.

In order to import a training set into PyJibe, the text files (train_feat_*.txt and train_response.txt) must be zipped and named according to the scheme, where NAME is a descriptive name.


The import in PyJibe is carried out by selecting Add… in the Scheme dropdown menu, located in the Curve rating box (lower right of a Force-Distance analysis). A dialog will ask for the zipped training set (e.g.

The import is persistent, i.e. the training set is extracted to the user’s configuration directory and thus does not have to be imported again the next time PyJibe is run.