Run single experiment with the code
python efficient-lda.py <lda_type> <K> <max_iter> <log name>
where
<lda_type>can be 'baseline', 'sparse' and 'alias'<K>is the number of topics<max_iter>designates the maximum iteration number<log name>is the path to log file. Time information and loglikelihood will be written into the log file.
Run all the experiments with the code
python shell.py
where shell.py implements all the combinations of different LDA algorithms and different LDA algorithms