tf.keras

1. 保存模型注意点
   When saving a model’s weights, tf.keras defaults to the checkpoint format. Pass save_format=‘h5’ to use HDF5 (or pass a filename that ends in .h5).
2. kernel开头的参数
   其表示的是w
3. fit
   用于训练，其返回值是一个history。其里面记录了训练的时候的一些参数
   所以其最后可以用于可视化

model = tf.keras.Sequential([
# Adds a densely-connected layer with 64 units to the model:
keras.layers.Dense(64, activation='relu', input_shape=(32,)),
# Add another:
keras.layers.Dense(64, activation='relu'),
# Add a softmax layer with 10 output units:
keras.layers.Dense(10, activation='softmax')])

model.compile(optimizer=tf.keras.optimizers.RMSprop(0.01),
              loss=tf.keras.losses.CategoricalCrossentropy(),
              metrics=[tf.keras.metrics.CategoricalAccuracy()])
#%%
data = np.random.random((1000, 32))
labels = np.random.random((1000, 10))

val_data = np.random.random((100, 32))
val_labels = np.random.random((100, 10))

history = model.fit(data, labels, epochs=10, batch_size=32,
          validation_data=(val_data, val_labels))

# 得到是一个字典
history.history.keys()

# 输出如下
dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy'])

4. model.fit
   也可以在训练的时候传入一个Dataset。这个Dataset里面的一个元素，不仅具有x，还具有y

# Instantiates a toy dataset instance:
dataset = tf.data.Dataset.from_tensor_slices((data, labels))
dataset = dataset.batch(32)

model.fit(dataset, epochs=10)

dataset = tf.data.Dataset.from_tensor_slices((data, labels))
dataset = dataset.batch(32)

val_dataset = tf.data.Dataset.from_tensor_slices((val_data, val_labels))
val_dataset = val_dataset.batch(32)

model.fit(dataset, epochs=10,
          validation_data=val_dataset)

# With a Dataset
dataset = tf.data.Dataset.from_tensor_slices((data, labels))
dataset = dataset.batch(32)

model.evaluate(dataset)