def reset(self):
if self.shuffle:
self.source, self.target = shuffle.main(
[self.source_orig, self.target_orig], temporary=True)
else:
self.source.seek(0)
self.target.seek(0)
def __init__(self, source, source_dict,
batch_size=128,
n_words_source=-1,
skip_empty=False,
shuffle_each_epoch=False,
sort_by_length=False,
maxibatch_size=20,
):
if shuffle_each_epoch:
self.source_orig = source
self.source = shuffle.main([self.source_orig], temporary=True)
else:
self.source = data_utils.fopen(source, 'r')
self.source_dict = load_dict(source_dict)
self.batch_size = batch_size
self.skip_empty = skip_empty
self.n_words_source = n_words_source
if self.n_words_source > 0:
for key, idx in self.source_dict.items():
if idx >= self.n_words_source:
del self.source_dict[key]
self.shuffle = shuffle_each_epoch
self.sort_by_length = sort_by_length
self.source_buffer = []
self.label_buffer =[]
self.k = batch_size * maxibatch_size
self.end_of_data = False
def __init__(self,
source,
target,
source_dict,
target_dict,
batch_size=128,
maxlen=100,
n_words_source=-1,
n_words_target=-1,
skip_empty=False,
shuffle_each_epoch=False,
sort_by_length=True,
maxibatch_size=20):
if shuffle_each_epoch:
self.source_orig = source
self.target_orig = target
self.source, self.target = shuffle.main(
[self.source_orig, self.target_orig], temporary=True)
else:
self.source = open(source, 'r', encoding='utf-8')
self.target = open(target, 'r', encoding='utf-8')
self.source_dict = load_dict(source_dict)
self.target_dict = load_dict(target_dict)
self.batch_size = batch_size
self.maxlen = maxlen
self.skip_empty = skip_empty
self.n_words_source = n_words_source
self.n_words_target = n_words_target
# todo: limit vocabulary size
if self.n_words_source > 0:
for key, idx in self.source_dict.items(): # add list(...)
if idx >= self.n_words_source:
del self.source_dict[key]
if self.n_words_target > 0:
for key, idx in self.target_dict.items():
if idx >= self.n_words_target:
del self.target_dict[key]
self.shuffle = shuffle_each_epoch
self.sort_by_length = sort_by_length
self.source_buffer = []
self.target_buffer = []
self.k = batch_size * maxibatch_size
self.end_of_data = False
def reset(self):
if self.shuffle:
self.source = shuffle.main([self.source_orig], temporary=True)
else:
self.source.seek(0)