I would like to know what all the abbreviations mean? Some I can guess, like "PUNCT", but no idea what "X" might be. I want to retain contractions, but hard to choose options without documentation.

Thanks. Great performance code!

Is there a way to get the topic mixture of each document back out from a hierarchical model? I am training a HLDAModel:

h_mdl = tp.HLDAModel(depth=4,corpus=corpus,seed=1)
    
for i in range(0, 100, 10): #Train the model using Gibbs-sampling
    h_mdl.train(10)
    print('Iteration: {}\tLog-likelihood: {}'.format(i, h_mdl.ll_per_word))

I am using the Document class to access insta