John Bauer
John Bauer
Hopefully you mean using all the code at 056c413, not just cherrypicking, since that particular change was just a refactoring to set up a tagger change. Thanks for checking! My...
What I just did instead is output the result of `SystemUtils.getMemoryInUse()`
hmm, that doesn't seem very reliable
If the goal is to avoid peak memory usage, interning the lexicon after loading it might not help
However, creating the lexicon with interned strings should do it, right? Just not this time around... would only help the next time we retrain & reserialize the models. I suppose...
That seems to work exactly as I had hoped (although the models don't get smaller on disk, sadly)
Dedup before serializing is quite a large fraction of the gain, and I'm a little adverse to changing the readObject method because of how confusing it becomes for people working...
Lots, for better or worse. It does make people's lives easier in the short term, especially when you can wrap it into a single function call as we have for...
Certainly this is not ideal, but I can also kind of see why it would happen. The city of Victoria is on Vancouver Island, and the location Victoria shows up...
If you can tell us what's going on with stanza, we can investigate that, but we're not responsible for the stanfordcorenlp package