Usage: words [-dhNo] [DATASETS...] [NUMBER_OF_WORDS]

options:
  -l, --list             list valid datasets
  -d, --debug            debugging output
  -N, --dont-normalize   don't normalize frequencies when combining
                         multiple Markov models; this has the effect
                         of making larger datasets more influential
  -o, --target-offset    change the target length offset used in the
                         word generation algorithm; use negative integers
                         for best results