def load_instances(instances_lines, word_vectors):
'''Load real training examples
Args:
instance_lines: each string is a training example
word_vectors: an instance of vec.wordvector
Return:
instances: a list of ReorderInstance
'''
instances = [ReorderInstance.paser_from_str(i, word_vectors) for i in instances_lines]
instances = [i for i in instances if len(i.preWords) != 0 and len(i.aftWords) != 0]
return instances
def prepare_data(word_vectors=None, dataFile=None, unlabelFile=None):
'''Prepare training data
Args:
word_vectors: an instance of vec.wordvector
dataFile: raw training file
Return:
instances: a list of ReorderInstance
word_vectors
'''
if rank == 0:
comm.bcast(word_vectors, root=0)
instance_of_domain = []
instance_lines = []
lines_of_Unlabel = []
instances_of_Unlabel = []
if unlabelFile != None:
with Reader(unlabelFile) as file:
for line in file:
lines_of_Unlabel.append(line)
instances_of_Unlabel = [ReorderInstance.paser_from_unlabeled_str(i, word_vectors) for i in lines_of_Unlabel]
instances_of_Unlabel = [i for i in instances_of_Unlabel if len(i.preWords) != 0 and len(i.aftWords) != 0]
comm.bcast(instances_of_Unlabel, root=0)
# if type(dataFile) == str:
# with Reader(dataFile) as file:
# for line in file:
# instance_of_domain.append(line)
# instances = load_instances(instance_of_domain, word_vectors)
# if unlabelFile != None:
# return instances, instances_of_Unlabel, word_vectors
# return instances, word_vectors
for file in dataFile:
with Reader(file) as file:
for line in file:
instance_of_domain.append(line)
instance_lines.append(instance_of_domain)
instance_of_domain = []
for i in range(1, worker_num):
comm.send(instance_lines[i], dest=i)
comm.barrier()
instances = load_instances(instance_lines[0], word_vectors)
del instance_lines
if unlabelFile != None:
return instances, instances_of_Unlabel, word_vectors
return instances, None, word_vectors
else:
word_vectors = comm.bcast(root=0)
instances_of_Unlabel = comm.bcast(root=0)
instances_lines = comm.recv(source=0)
comm.barrier()
instances = load_instances(instances_lines, word_vectors)
if unlabelFile != None:
return instances, instances_of_Unlabel, word_vectors
return instances, None, word_vectors
