If the implementation is to be found, it's worth being able to produce a ChronicleMap version of word2vec model. It has much lower memory footprint and the access is speed is enough for most experiments.
It also plays nicely with Spark, so more complicated stuff with vectors can be expressed more efficiently.
Depends on #25
If the implementation is to be found, it's worth being able to produce a ChronicleMap version of word2vec model. It has much lower memory footprint and the access is speed is enough for most experiments.
It also plays nicely with Spark, so more complicated stuff with vectors can be expressed more efficiently.