I have been using the mallet
package in R that wraps MALLET. So far so good, but I was wondering how I can save and load th model state on the disk, so that I don't have to train all over and get a different model each time.
For the saving part it seems that model$writeState()
does the trick, although this function is a bit undocumented. The complete code for saving the model state would be something like
model <- trainLDA(mallet.instances, 30)
model$writeState('model.state.gz')
Loading still remains a mystery to me. Perhaps calling a method from the base class ParallelTopicModel
but don't really now how to do it in rJava