Ask HN: Word Embedding / Word2Vec for Locations?

Wondering if anyone has come across a word embedding type dataset for high cardinality geography data? I imagine (for the US) embeddings for states, zip codes and counties, to capture similarities across a multitude of dimensions.