Keras高层API
基本就是4步：
Matrics
update_state
result().numpy()
reset_states(就是清除缓存)
tensorflow2.0代码很简单，就MNIST数据集演示：
import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics
def preprocess(x, y):
"""
x is a simple image, not a batch
"""
x = tf.cast(x, dtype=tf.float32) / 255.
x = tf.reshape(x, [28 * 28])
y = tf.cast(y, dtype=tf.int32)
y = tf.one_hot(y, depth=10)
return x, y
batchsz = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())
db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(preprocess).shuffle(60000).batch(batchsz)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(preprocess).batch(batchsz)
sample = next(iter(db))
print(sample[0].shape, sample[1].shape)
network = Sequential([layers.Dense(256, activation='relu'),
layers.Dense(128, activation='relu'),
layers.Dense(64, activation='relu'),
layers.Dense(32, activation='relu'),
layers.Dense(10)])
network.build(input_shape=(None, 28 * 28))
network.summary()
network.compile(optimizer=optimizers.Adam(lr=0.001),
loss=tf.losses.CategoricalCrossentropy(from_logits=True),
metrics=['accuracy']
)
network.fit(db, epochs=5, validation_data=ds_val, validation_freq=1)
network.evaluate(ds_val)
sample = next(iter(ds_val))
x = sample[0]
y = sample[1] # one-hot
pred = network.predict(x) # [b, 10]
# convert back to number
y = tf.argmax(y, axis=1)
pred = tf.argmax(pred, axis=1)
print(pred)
print(y)
结果：
Epoch 5/5
469/469 [==============================] - 2s 4ms/step - loss: 0.0850 - accuracy: 0.9769 - val_loss: 0.1236 - val_accuracy: 0.9688
79/79 [==============================] - 1s 9ms/step - loss: 0.1236 - accuracy: 0.9688
tf.Tensor(
[7 2 1 0 4 1 4 9 6 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 2 7
1 2 1 1 7 4 4 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 7 0 2 9
1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 4 3 1 4 1 7 6 9 6 0 5 4 9 9 2 1 9 4 8
7 3 9 7 4 4 4 9 2 5 4 7 6 7 9 0 5], shape=(128,), dtype=int64)
tf.Tensor(
[7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 2 7
1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 7 0 2 9
1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9 6 0 5 4 9 9 2 1 9 4 8
7 3 9 7 4 4 4 9 2 5 4 7 6 7 9 0 5], shape=(128,), dtype=int64)
CIFAR10实战，利用自定义层处理
import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics
from tensorflow import keras
def preprocess(x, y):
# 范围转换从[0,255] => [-1,1]
x = 2 * tf.cast(x, dtype=tf.float32) / 255.0 - 1.0
y = tf.cast(y, dtype=tf.float32)
return x, y
batchsz = 128
# [50000,32,32,3], [50000,1,10]
(x, y), (x_val, y_val) = datasets.cifar10.load_data()
# 默认删除所有为1的维度
y = tf.squeeze(y)
y_val = tf.squeeze(y_val)
y = tf.one_hot(y, depth=10) # [50000, 10]
y_val = tf.one_hot(y_val, depth=10) # [10000, 10]
print("datasets:", x.shape, y.shape, x_val.shape, y_val.shape)
train_db = tf.data.Dataset.from_tensor_slices((x, y))
train_db = train_db.map(preprocess).shuffle(10000).batch(batchsz)
test_db = tf.data.Dataset.from_tensor_slices((x_val, y_val))
test_db = test_db.map(preprocess).batch(batchsz)
sample = next(iter(train_db))
print(sample[0].shape, sample[1].shape)
class MyDense(layers.Layer):
# 代替了标准的layers.Dense()
def __init__(self, inp_dim, out_dim):
super(MyDense, self).__init__()
self.kernel = self.add_weight('w', [inp_dim, out_dim])
# self.bias = self.add_variable('b', [out_dim])
def call(self, inputs, training=None):
x = inputs @ self.kernel
return x
class MyNetwork(keras.Model):
def __init__(self):
super(MyNetwork, self).__init__()
self.fc1 = MyDense(32 * 32 * 3, 256)
self.fc2 = MyDense(256, 128)
self.fc3 = MyDense(128, 64)
self.fc4 = MyDense(64, 32)
self.fc5 = MyDense(32, 10)
def call(self, inputs, training=None):
'''
:param inputs: [b, 32, 32, 3]
:param training:
:return:
'''
x = tf.reshape(inputs, [-1, 32 * 32 * 3])
# 转换维度, [b, 32*32*3] => [b, 256]
x = self.fc1(x)
x = tf.nn.relu(x)
# 转换维度, [b, 256] => [b, 128]
x = self.fc2(x)
x = tf.nn.relu(x)
# 转换维度, [b, 128] => [b, 64]
x = self.fc3(x)
x = tf.nn.relu(x)
# 转换维度, [b, 64] => [b, 32]
x = self.fc4(x)
x = tf.nn.relu(x)
# 转换维度, [b, 32] => [b, 10]
x = self.fc5(x)
return x
network = MyNetwork()
network.compile(optimizer=optimizers.Adam(lr=1e-3),
loss=tf.losses.CategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
network.fit(train_db, epochs=15, validation_data=test_db, validation_freq=1)
network.evaluate(test_db)
network.save_weights('ckpt/weights.ckpt')
del network
print("saved to ckpt/weights.ckpt")
network = MyNetwork()
network.compile(optimizer=optimizers.Adam(lr=1e-3),
loss=tf.losses.CategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
network.load_weights('ckpt/weights.ckpt')
print("loaded weights from files")
network.evaluate(test_db)
结果：
Epoch 15/15
391/391 [==============================] - 2s 6ms/step - loss: 0.7328 - accuracy: 0.7411 - val_loss: 1.7053 - val_accuracy: 0.5153
79/79 [==============================] - 0s 4ms/step - loss: 1.7053 - accuracy: 0.5153
saved to ckpt/weights.ckpt
loaded weights from files
79/79 [==============================] - 0s 4ms/step - loss: 1.7053 - accuracy: 0.5153
模型的保存和加载
保存权值
上个cifar10例子就有保存权值的例子。
保存整个模型（包括权值）
import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics
def preprocess(x, y):
"""
x is a simple image, not a batch
"""
x = tf.cast(x, dtype=tf.float32) / 255.
x = tf.reshape(x, [28*28])
y = tf.cast(y, dtype=tf.int32)
y = tf.one_hot(y, depth=10)
return x,y
batchsz = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())
db = tf.data.Dataset.from_tensor_slices((x,y))
db = db.map(preprocess).shuffle(60000).batch(batchsz)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(preprocess).batch(batchsz)
sample = next(iter(db))
print(sample[0].shape, sample[1].shape)
network = Sequential([layers.Dense(256, activation='relu'),
layers.Dense(128, activation='relu'),
layers.Dense(64, activation='relu'),
layers.Dense(32, activation='relu'),
layers.Dense(10)])
network.build(input_shape=(None, 28*28))
network.summary()
network.compile(optimizer=optimizers.Adam(lr=0.01),
loss=tf.losses.CategoricalCrossentropy(from_logits=True),
metrics=['accuracy']
)
network.fit(db, epochs=3, validation_data=ds_val, validation_freq=2)
network.evaluate(ds_val)
network.save('model.h5')
print('saved total model.')
del network
print('loaded model from file.')
network = tf.keras.models.load_model('model.h5', compile=False)
network.compile(optimizer=optimizers.Adam(lr=0.01),
loss=tf.losses.CategoricalCrossentropy(from_logits=True),
metrics=['accuracy']
)
x_val = tf.cast(x_val, dtype=tf.float32) / 255.
x_val = tf.reshape(x_val, [-1, 28*28])
y_val = tf.cast(y_val, dtype=tf.int32)
y_val = tf.one_hot(y_val, depth=10)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val)).batch(128)
network.evaluate(ds_val)
其实都是差不多的，只是再加载时，不用创建网络模型了，加载时直接加载整个模型和权值，然后再直接训练，虽然可以完整保存和加载，但占用的存储空间很大，加载时也很慢，所以我们一般只保存权值参数，网络模型一般是不变的，所以不用保存。
本站所有文章均为原创，欢迎转载，请注明文章出处：https://blog.csdn.net/weixin_45092662。百度和各类采集站皆不可信，搜索请谨慎鉴别。技术类文章一般都有时效性，本人习惯不定期对自己的博文进行修正和更新，因此请访问出处以查看本文的最新版本。
1224
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
