f0 = model.Wf(x_k) + model.Rf(h)
f1 = F.sigmoid(f0)
c = i1 * z1 + f1 * c
o0 = model.Wo(x_k) + model.Ro(h)
o1 = F.sigmoid(o0)
y = o1 * F.tanh(c)
yv = F.softmax(model.W(y))
pi = yv.data[0][w2]
sum -= math.log(pi, 2)
return sum
model_filename = sys.argv[1]
vocab = {}
train_data = helpers.load_data('.data/ptb.train.min.txt', vocab)
eos_id = vocab['<eos>']
max_id = len(vocab) - 1
demb = 100
model = with_dropout.Lstm(len(vocab), eos_id, demb)
serializers.load_npz(model_filename, model)
test_data = helpers.load_data('.data/ptb.test.txt', vocab)
test_data = test_data[0:1000]
s = []
has_unknown = False
total_word_num = 0
sum = 0.0
for i in range(1, len(s)):
w1, w2 = s[i - 1], s[i]
x_k = model.embed(
Variable(np.array([w1], dtype=np.int32), volatile='on'))
y = model.H(x_k)
h = y
yv = F.softmax(model.W(y))
pi = yv.data[0][w2]
sum -= math.log(pi, 2)
return sum
model_filename = sys.argv[1]
vocab = {}
train_data = helpers.load_data('.data/ptb.train.min.txt', vocab)
eos_id = vocab['<eos>']
max_id = len(vocab) - 1
demb = 100
model = use_lstm.Lstm(len(vocab), eos_id, demb)
serializers.load_npz(model_filename, model)
test_data = helpers.load_data('.data/ptb.test.txt', vocab)
test_data = test_data[0:1000]
s = []
has_unknown = False
total_word_num = 0
sum = 0.0
import numpy as np
import chainer
from chainer import cuda, Function, gradient_check, Variable, optimizers, serializers, utils
from chainer import Link, Chain, ChainList
import chainer.functions as F
import chainer.links as L
from impls import use_lstm, helpers
name = helpers.get_script_name(__file__)
data_path = '.data/ptb.train.min.txt'
dest_root = '.dest'
vocab = {}
train_data = helpers.load_data(data_path, vocab)
vocab_num = len(vocab)
eos_id = vocab['<eos>']
demb = 100
model = use_lstm.Lstm(vocab_num, eos_id, demb)
optimizer = optimizers.Adam()
optimizer.setup(model)
for epoch in range(5):
s = []
for pos in range(len(train_data)):
id = train_data[pos]
s.append(id)
if id == eos_id:
