Why is 1GB of ram required to load a 160mb word vector file?

Are there some pre-computations done which cause this lib to eat up RAM? Because I'm using a 160mb text file of word vectors and the node process is taking up 900mb+ of RAM.

Just wondering whether there is a good reason for this, or whether I should dig about looking for some inefficiencies somewhere.

Thanks
