sotastream package

Subpackages

Submodules

Module contents

class sotastream.Defaults[source]

Bases: object

Default values for pipeline arguments

BUFFER_SIZE = 1000000
DOC_PROB = 0.0
DOC_PROB_PARALLEL = 0.0
DOC_SEPARATOR = ' <eos>'
MAX_TOKENS = 250
NUM_PROCESSES = 16
QUEUE_BUFFER_SIZE = 10000
QUIET = False
SAMPLE_FILE = None
SAMPLE_LENGTH = True
SEED = 0
SEPARATOR = ' '
SHUFFLE = True
SPM_MODEL = None