public class InMemoryCollapsedVariationalBayes0
Runs the same algorithm as CVB0Driver, but sequentially, in memory. Memory requirements
are currently: the entire corpus is read into RAM, two copies of the model (each of size
numTerms * numTopics), and another matrix of size numDocs * numTopics is held in memory
(to store p(topic|doc) for all docs).
But if all this fits in memory, this can be significantly faster than an iterative MR job.
Fields inherited from class org.apache.mahout.common.AbstractJob