def report(docs, t, idx, loss, results):
print "{0} epoch, {1}th data".format(t, idx)
print loss.data, np.linalg.norm(cuda.to_cpu_async(model.Wh1.W)), np.linalg.norm(
cuda.to_cpu_async(model.Wh2.W)
), np.linalg.norm(cuda.to_cpu_async(model.Wx1.W))
print "------------------------------------------"
print to_text([d[0] for d in docs], vocablary)
print "=========================================="
print to_text(results, vocablary)
print "------------------------------------------"
def report(docs, t, idx, loss, results):
print "{0} epoch, {1}th data".format(t, idx)
print loss.data, np.linalg.norm(cuda.to_cpu_async(
model.Wh1.W)), np.linalg.norm(cuda.to_cpu_async(
model.Wh2.W)), np.linalg.norm(cuda.to_cpu_async(model.Wx1.W))
print "------------------------------------------"
print to_text([d[0] for d in docs], vocablary)
print "=========================================="
print to_text(results, vocablary)
print "------------------------------------------"
def forward_one_step(h1, h2, cur_word, next_word, volatile=False):
word = V(cur_word, volatile=volatile)
x = F.leaky_relu(model.embed(word))
tmp_x = model.Wx1(x)
tmp_h1 = model.Wh1(h1)
h1 = F.leaky_relu(tmp_x + tmp_h1)
tmp_x2 = model.Wx2(h1)
tmp_h2 = model.Wh2(h2)
h2 = F.leaky_relu(tmp_x2 + tmp_h2)
y = model.Wy(h2)
t = V(next_word, volatile=volatile)
loss = F.softmax_cross_entropy(y, t)
pred = F.softmax(y)
return h1, h2, loss, np.argmax(cuda.to_cpu_async(pred.data))
def forward_one_step(h1, h2, cur_word, next_word, volatile=False):
word = V(cur_word, volatile=volatile)
x = F.leaky_relu(model.embed(word))
tmp_x = model.Wx1(x)
tmp_h1 = model.Wh1(h1)
h1 = F.leaky_relu(tmp_x + tmp_h1)
tmp_x2 = model.Wx2(h1)
tmp_h2 = model.Wh2(h2)
h2 = F.leaky_relu(tmp_x2 + tmp_h2)
y = model.Wy(h2)
t = V(next_word, volatile=volatile)
loss = F.softmax_cross_entropy(y, t)
pred = F.softmax(y)
return h1, h2, loss, np.argmax(cuda.to_cpu_async(pred.data))
def serialize_model(namespace, epoch, model, vocab):
file_path = "data/model_{0}_{1}.pickle".format(namespace, epoch)
pickle.dump((cuda.to_cpu_async(model), vocab), open(file_path, "wb"), -1)
def serialize_model(namespace, epoch, model, vocab):
file_path = "data/model_{0}_{1}.pickle".format(namespace, epoch)
pickle.dump((cuda.to_cpu_async(model), vocab), open(file_path, 'wb'), -1)
