Github user KellenSunderland commented on the issue:
This PR should now implement all of the JNI improvements we wanted to make
at the MT Marathon. Before merging would you be able to run a quick test and
see what the performance is like? If the second commit doesn't speed things up
we could consider removing it to reduce code complexity.
I also noticed that language model ffs all were estimating with the
`public float estimateCost(Rule rule, Sentence sentence)`
Does it actually make sense to pass the sentence object here? Each rule
should be able to be estimated with only its local feature score + weights
right? It shouldn't need any additional context?
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket