pseudo-random text generator using ngram models
Generate text using randomized word ngrams from standard input.
positional arguments:
n the number of words to consider simultaneously when
generating text.
min_length each iteration will be retried until the minimum
length (in words) is reached.
max_length length (in words) at which to cease generating text if
no end token found. Note: the [min, max] length range
is inclusive.
iterations The number of text snippets to generate.
optional arguments:
-h, --help show this help message and exit
--text_files TEXT_FILES [TEXT_FILES ...]
optional list of files to read in lieu of stdin.
--start_token START_TOKEN
predetermined first word of output text
--end_token END_TOKEN
delimeter at which a given text generation iteration
stops; not included in generated text. Must be
distinct from the start token.
--seed SEED seed to control random text generation
--ngram_model_file NGRAM_MODEL_FILE
if a file path ending in .ngrams is specified, the
ngram model will be deserialized FROM the provided
path (if it exists); otherwise, it will be treated as
an output parameter, and the model will be serialized
TO the provided path.