L-BFGS chose a non-descent direction

[ericleasemorgan]

Using the following command line, I am trying to re-create the out-of-the box BookNLP model:

./runjava novels.training/TrainCoref -training coref/annotatedData.txt -o coref/weights.txt

Unfortunately, processing does not complete, but instead returns the following error:

L1 = 2.0, L2 = 0.1
regressing with 0 training samples of length 0
Exception in thread "main" java.lang.RuntimeException: L-BFGS chose a non-descent direction: check your gradient!
	at novels.optimization.OptimizerState.backTrackingLineSearch(OWLQN.java:410)
	at novels.optimization.OWLQN.minimize(OWLQN.java:147)
	at novels.optimization.SparseRegression.regress(SparseRegression.java:39)
	at novels.training.TrainCoref.train(TrainCoref.java:170)
	at novels.training.TrainCoref.main(TrainCoref.java:244)

How do I denote the number of training samples, and how do I configure the value of L-BFGS?