indices = np.arange(len(inputs))
np.random.shuffle(indices)
for start_idx in range(0, len(inputs) - batch_size + 1, batch_size):
if shuffle:
excerpt = indices[start_idx:start_idx + batch_size]
else:
excerpt = slice(start_idx, start_idx + batch_size)
yield inputs[excerpt], targets[excerpt]
x = tf.placeholder(tf.float32, shape=[None, w, h, c], name='x')
y_ = tf.placeholder(tf.int32, shape=[
None,
], name='y_')
logits = CNN.CNNlayer(x)
loss = tf.losses.sparse_softmax_cross_entropy(labels=y_, logits=logits)
train_op = tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
correct_prediction = tf.equal(tf.cast(tf.argmax(logits, 1), tf.int32), y_)
acc = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
# 训练和测试数据
saver = tf.train.Saver(max_to_keep=3)
max_acc = 0
f = open('train/acc.txt', 'w')
n_epoch = 10
batch_size = 64
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
for epoch in range(n_epoch):
#
# test_data_length = [len(item) for item in test_x]
# max_length = np.max(test_data_length)
# print(max_length)
dev_x_tensor = data_loader.pad(dev_x, word2id, "-pad-",38)
dev_y_tensor = torch.from_numpy(np.array(dev_y))
test_x_tensor = data_loader.pad(test_x, word2id, "-pad-",38)
test_y_tensor = torch.from_numpy(np.array(test_y))
# print(test_x_tensor)
plot_dev = []
plot_test = []
x = []
model = CNN.CNNlayer(len(word2id), 100, 2, [3, 4, 5],output_size=2)
with torch.no_grad():
for i in range(100):
# model = pooling.Pooling(len(word2id), 100, 2)
model.eval()
model.load_state_dict(torch.load(model_url+"/"+str(i+1)+".pt"))
correct = 0
total = 0
outputs1 = model(dev_x_tensor.long())
dev_right_num = train.accuracy_num(outputs1, dev_y_tensor)
dev_acc = float(dev_right_num)/len(dev_y_tensor)*100
outputs2 = model(test_x_tensor.long())
test_right_num = train.accuracy_num(outputs2, test_y_tensor)
test_acc = float(test_right_num )/ len(test_y_tensor) * 100
plot_dev.append(dev_acc)
