当我创建一个简单的Keras模型时
model = Sequential()
model.add(Dense(10, activation='tanh', input_dim=1))
model.add(Dense(1, activation='linear'))
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mean_squared_error'])
Run Code Online (Sandbox Code Playgroud)
并回调到Tensorboard
tensorboard = TensorBoard(log_dir='c:/temp/tensorboard/run1', histogram_freq=1, write_graph=True, write_images=False)
model.fit(x, y, epochs=1000, batch_size=1, callbacks=[tensorboard])
Run Code Online (Sandbox Code Playgroud)
换句话说,这是一团糟。
小智 5
您可以使用来创建名称范围以将模型中的图层分组K.name_scope('name_scope')。
例:
with K.name_scope('CustomLayer'):
# add first layer in new scope
x = GlobalAveragePooling2D()(x)
# add a second fully connected layer
x = Dense(1024, activation='relu')(x)
Run Code Online (Sandbox Code Playgroud)
感谢 https://github.com/fchollet/keras/pull/4233#issuecomment-316954784