train_embed() cannot update model with corpus_file

[hannahxchen]

Test code:

from embedding import SecWord2Vec

keywords = ['hello', 'sentence', 'meow']
model = SecWord2Vec(keywords, corpus_file='data/test.txt', min_count=1, size=10, iter=1)
model.train_embed()

print('corpus_file_1:')
for line in open('data/test.txt'):
	print(line.strip())

print('\nBefore update')
print('vocabs: {}'.format(model.wv.vocab.keys()))
print('keyword corpus: {}'.format(model.kc))
print('sentences:{}'.format(model.sentences))

print('\ncorpus_file_2:')
for line in open('data/test_2.txt'):
	print(line.strip())

model.train_embed(corpus_file='data/test_2.txt')
print('\nAfter update')
print('vocabs: {}'.format(model.wv.vocab.keys()))
print('keyword corpus: {}'.format(model.kc))
print('sentences:{}'.format(model.sentences))

Output:

[gen3111620]

Solution:

at train_embed function add code

test_code:

w2v = SecGloVe(keywords, corpus_file=corpus_file, size=10, window=5, min_count=1, 
	case_sensitive=True, iter=10, corpus_chunksize=20000, corpus_worker=5)
# print(w2v.wv.vocab.keys())
w2v.train_embed()

print(w2v.wv['hello'])
w2v.train_embed(corpus_file=new_corpus_file)
print(w2v.wv['hello'])

out_put:

** keywords update maybe need test too.