Hi, I'm running the new implementation of LDA algorithm. At first, I tried with 75 topic extraction and then, I tried with 5 topics. The "problem" is that if I make the mean of all the positions in this matrix, the result is the same for those 2 executions. I dont know how to interpret this because, if I want 75 topics insted of 5, there should be more terms differents than 0 in the matrix, isn't it?
I wanna do the mean for aproximate the number of topics that I need, in order to see when the matrix dispersion is low (the mean is high) and use this number of topics for running LDA.
