def iter_XY(config):
(embeddingconfig, gramconfig, runconfig) = config
embedding = WordEmbedding(embeddingpath(embeddingconfig))
unknown_embedding = None
if not runconfig.filter_unknown:
unknown_embedding = zeros(embedding.values.shape[1])
def embed(gram):
return embedding.embed(gram, unknown_embedding)
with gzip.open(grampath(gramconfig), mode='rt') as f:
for line in f:
line = line.split()
if len(line) == gramconfig.gram_size:
gram = line
lbl = word_to_label(gramconfig.skipwords, '')
elif len(line) != gramconfig.gram_size + 1:
print('Skip length > 1 not supported. Dropping Gram:', *line, file=stderr)
continue
else:
gram = line[ : gramconfig.skippos] + line[gramconfig.skippos + 1 : ]
skip = line[gramconfig.skippos]
lbl = word_to_label(gramconfig.skipwords, skip)
try:
gramvec = embed(gram)
yield gramvec, lbl
except KeyError:
pass
def main():
embedding = WordEmbedding(embeddingpath(default_embeddingconfig))
for old, new in spelling_changes:
print(old, '--', new)
print(embedding.nearest_words([old]))
print()
print()
war, ist = tense_changes[0]
tensediff = embedding[ist] - embedding[war]
for past, present in tense_changes[1 : ]:
print(past, '+ tensediff:', *embedding.nearest_words([embedding[past] + tensediff]))
print('Should be:', present)
print()
# word_diffs = [embedding[new] - embedding[old] for (old, new) in word_changes]
spelling_diffs = [embedding[new] - embedding[old] for (old, new) in spelling_changes[10 : 20]]
tense_diffs = [embedding[present] - embedding[past] for (past, present) in tense_changes]
def metric(u, v):
return max(distance.cosine(u, v), 0)
while True:
try:
model = TSNE(n_components=2, metric=metric)
reduced = model.fit_transform(spelling_diffs + tense_diffs)
print(reduced)
return
except Exception:
pass
def create_embedding(embeddingconfig):
makedirs('data/embeddings', exist_ok=True)
corpuspath = config.corpuspath(embeddingconfig.corpus)
embeddingpath = config.embeddingpath(embeddingconfig)
if path.isfile(embeddingpath):
return
create_word_embedding( infile=corpuspath \
, outfile=embeddingpath \
, size=embeddingconfig.dimension \
, estimator=embeddingconfig.estimator \
, negative=embeddingconfig.negative \
, downsample=embeddingconfig.downsampling \
, min_count=embeddingconfig.min_count )