Using multiprocessing in python to speed up the storing process in a dictionary

Question

I have a dictionary of 100k keys with their glove embeddings as values. Now I want to find the top 10 elements in the dictionary related to a element on the basis of its cos