For implementing ER model in Saul-SL, it would be much more compact to stick to the same existing EntityRelationClassifiers declarations. However, for SL, I use SparseNetworks rather than SparsePerceptrons. By changing EntityRelationClassifiers types, the ER Test Units fail to Load the trained SparsePerceptron models. So, it would be great if we train SparseNetworks also for ER and have them in the ER models Jar.
For implementing ER model in Saul-SL, it would be much more compact to stick to the same existing EntityRelationClassifiers declarations. However, for SL, I use SparseNetworks rather than SparsePerceptrons. By changing EntityRelationClassifiers types, the ER Test Units fail to Load the trained SparsePerceptron models. So, it would be great if we train SparseNetworks also for ER and have them in the ER models Jar.