def GET(self, arg=None):
# Get query params
query = web.input(
bits = 6,
words = 5
)
# Change output to JSON
web.header('Content-type', 'application/json')
# If no pattern at the end of the url,
# we will generate a random password
if not arg:
try:
words = model.get_words(results = 2**int(query.bits))
# Convert iterator
wordlist = []
for word in words:
wordlist.append(word.word)
except(), e:
web.internalerror(str(e))
raise
try:
generatedPass = generate_password(
int(query.words), wordlist
)
except(), e:
web.internalerror(str(e))
raise
def generate():
string = request.form.get('text')
number = request.form['nextstep']
if number is '': number = 35
if string is '': string = 'Oh Romeo, Oh Romeo,'
prediction = get_words(model, string, int(number))
result = "Result"
return render_template('index.html', prediction=prediction, result=result)
def corpus_words_list():
word_list = model.get_words()
if session:
logged_in = 'Logged in as: %s.'%session['user']
not_you = 'Not %s?'%session['user']
return render_template('browse_list_words.html', logged_in=logged_in, corpus_list=word_list, not_you=not_you)
else:
return render_template('browse_list_words.html', corpus_list=word_list)
def corpus_words_list():
word_list = model.get_words()
if session:
logged_in = 'Logged in as: %s.' % session['user']
not_you = 'Not %s?' % session['user']
return render_template('browse_list_words.html',
logged_in=logged_in,
corpus_list=word_list,
not_you=not_you)
else:
return render_template('browse_list_words.html', corpus_list=word_list)
def main(data_dir, out_dir, n_iter=10, vector_len=300, vocab_size=20000,
hidden_len=300, depth=3, drop_rate=0.3, rho=1e-4, batch_size=24):
print("Loading")
nlp = spacy.en.English(parser=False)
dataset = Dataset(nlp, data_dir / 'train', batch_size)
print("Training")
network = model.train(dataset, vector_len, hidden_len, 2, vocab_size, depth,
drop_rate, rho, n_iter,
model_writer(out_dir, 'model_{epoch}.pickle'))
score = model.Scorer()
print("Evaluating")
for doc, label in read_data(nlp, data_dir / 'test'):
word_ids, embeddings = model.get_words(doc, 0.0, vocab_size)
guess = network.forward(word_ids, embeddings)
score += guess == label
print(score)
