Comment by kick_in_the_dor

OP has a point. Are these type of embeddings the best way to model thought?