train_data_iter = NDArrayIter(train_X,
train_Y,
batch_size=batch_size,
shuffle=True)
test_data_iter = NDArrayIter(test_X,
test_Y,
batch_size=batch_size,
shuffle=False)
model = LMModel(vocab_size=vocab_size, H_DIM=200, EMB_DIM=200)
updater = Updater(model, update_rule='sgd', learning_rate=0.001)
mt = time.time()
iter_num = 0
for ep in xrange(50):
train_data_iter.reset()
for batch in train_data_iter:
iter_num += 1
data, labels = unpack_batch(batch)
loss = model(data, labels=labels)
#grad_dict, loss = model.grad_and_loss(data, labels)
grad_dict = model.backward()
updater(grad_dict)
if iter_num % 10 == 0:
print 'Iteration {0:} Time {1:.2f}, loss {2:.4f}'.format(
iter_num,
time.time() - mt, float(loss.asnumpy()))
mt = time.time()
shuffle=False)
if args.rnn == 'RNN': model = RNNModel(128)
elif args.rnn == 'LSTM': model = LSTMModel(128)
updater = Updater(model, update_rule='rmsprop', learning_rate=0.002)
iteration_number = 0
for epoch_number in range(50):
for iteration, batch in enumerate(train_data_iter):
iteration_number += 1
data, labels = unpack_batch(batch)
grad_dict, loss = model.grad_and_loss(data, labels)
updater(grad_dict)
if iteration_number % 100 == 0:
print 'iteration %d loss %f' % (iteration_number, loss)
test_data_iter.reset()
errors, samples = 0, 0
for batch in test_data_iter:
data, labels = unpack_batch(batch)
scores = model.forward(data, 'inference')
predictions = np.argmax(scores, axis=1)
errors += np.count_nonzero(predictions - labels)
samples += data.shape[0]
print 'epoch %d validation error %f' % (epoch_number,
errors / float(samples))
train_data_iter = NDArrayIter(train_data, data['train_label'][:sample_number], batch_size, shuffle=True)
test_data_iter = NDArrayIter(test_data, data['test_label'][:sample_number], batch_size, shuffle=False)
if args.rnn == 'RNN': model = RNNModel(128)
elif args.rnn == 'LSTM' : model = LSTMModel(128)
updater = Updater(model, update_rule='rmsprop', learning_rate=0.002)
iteration_number = 0
for epoch_number in range(50):
for iteration, batch in enumerate(train_data_iter):
iteration_number += 1
data, labels = unpack_batch(batch)
grad_dict, loss = model.grad_and_loss(data, labels)
updater(grad_dict)
if iteration_number % 100 == 0:
print 'iteration %d loss %f' % (iteration_number, loss)
test_data_iter.reset()
errors, samples = 0, 0
for batch in test_data_iter:
data, labels = unpack_batch(batch)
scores = model.forward(data, 'inference')
predictions = np.argmax(scores, axis=1)
errors += np.count_nonzero(predictions - labels)
samples += data.shape[0]
print 'epoch %d validation error %f' % (epoch_number, errors / float(samples))