I noticed this (fairly old) package was throwing a lot of warnings when run in v0.4.5 and, moreover, had an horrible performance. It took more than two minutes for training a two-class SVM with 100 instances, each one composed of less than 700 floats. On the same input data, the svm-train CLI command took less than 0.2 secs.
Fixing the warnings fixed also this performance problem.
I noticed this (fairly old) package was throwing a lot of warnings when run in v0.4.5 and, moreover, had an horrible performance. It took more than two minutes for training a two-class SVM with 100 instances, each one composed of less than 700 floats. On the same input data, the
svm-train
CLI command took less than 0.2 secs.Fixing the warnings fixed also this performance problem.