Current trick for handling duplicates can lead to trouble with manual re-init

GateNLP / gateplugin-LearningFramework

A plugin for the GATE language technology framework for training and using machine learning models. Currently supports Mallet (MaxEnt, NaiveBayes, CRF and others), LibSVM, Scikit-Learn, Weka, and DNNs through Pytorch and Keras.

https://gatenlp.github.io/gateplugin-LearningFramework/

GNU Lesser General Public License v2.1

26 stars 6 forks source link

Current trick for handling duplicates can lead to trouble with manual re-init #106

Closed johann-petrak closed 5 years ago

johann-petrak commented 5 years ago

The current way of how each duplicate gets its own duplication number is based on a global variable getting incrementing when each duplicate gets initialised. Then, training is only performed by the duplicate which was the first.

This will cause the PR to fail if inside the GUI the user manually re-initialises the PR, thus incrementing the counter. We need to move the counter incrementing into custom duplication code.

ianroberts commented 5 years ago

Alternatively just add code in reInit() to clear the counter, so that a manual re-init is treated the same as the initial init rather than the same as a duplication.

johann-petrak commented 5 years ago

Current implementation uses custom duplication to do this, fixed by 0147dec62f6c694a81180965c09eccfb530238da