org.allenai.nlpstack.parse.poly.ml
the pools in the corpus
the pools in the corpus
Creates a corpus of "difference vectors" by repeatedly sampling two vectors from a candidate pool and subtracting them.
Creates a corpus of "difference vectors" by repeatedly sampling two vectors from a candidate pool and subtracting them. The resulting vector will have a negative cost if the first vector is better (has lower cost) than the second, and a positive cost if the second vector is better (has lower cost) than the first.
number of samples to generate per pool
a training data corpus consisting of the resulting difference vectors
A CandidatePoolCorpus is a set of candidate pools (see CandidatePool, above).
the pools in the corpus