I created an LDA visualization using pyLDAvis and was wondering what "token" means in the title of the bar graph (i.e., Topic 3 (14% of tokens)). I read the pyLDAvis documentation, but could not find an explanation. Does this mean that 14% of all the documents in the corpus fit into this topic or does it have to do with the distribution of words?

Thank you for the help.

Example bar graph that I generated using pyLDAvis