在使用全连接层的时候我们可以使用高级接口直接调用Dense层,那有没有想过自己实现一个Dense层的,下面我们一起实现一个Dense层
这是自己实现的第一个类,自动调用 call 方法,继承 layers.Layer 类,实现 y = wx + b class MyDense(layers.Layer): def __init__(self, inp_dim, outp_dim): # 这是输入维度 和 输出维度 super(MyDense, self).__init__() # 需要继承父类 self.kernel = self.add_variable('w', [inp_dim, outp_dim]) self.bias = self.add_variable('b', [outp_dim]) def call(self, inputs, **kwargs): out = inputs @ self.kernel + self.bias return out 接下来就是实现自己的模型 MyModel, 继承 keras.Model 类, 同样也是需要调用 super 方法 class MyModel(keras.Model): def __init__(self): super(MyModel, self).__init__() self.fc1 = MyDense(28 * 28, 256) self.fc2 = MyDense(256, 128) self.fc3 = MyDense(128, 64) self.fc4 = MyDense(64, 32) # 因为这个层是用来对 minist 数据集进行训练 所以最后的输出维度是 10 self.fc5 = MyDense(32, 10) def call(self, inputs, training=None, mask=None): x = self.fc1(inputs) # 中间加入 relu 非线性作为激活函数 x = tf.nn.relu(x) x = self.fc2(x) x = tf.nn.relu(x) x = self.fc3(x) x = tf.nn.relu(x) x = self.fc4(x) x = tf.nn.relu(x) x = self.fc5(x) return x 接下来就是数据预处理函数 def process(x, y): x = tf.cast(x, dtype=tf.float32) / 255. y = tf.cast(y, dtype=tf.int64) x = tf.reshape(x, [28 * 28]) y = tf.one_hot(y, depth=10) return x, y 训练数据部分 # 设置 batch_size 的大小 batch_size = 128 # 加载数据集 (x, y), (x_test, y_test) = datasets.mnist.load_data() db = tf.data.Dataset.from_tensor_slices((x, y)) db = db.map(process).shuffle(1000).batch(batch_size) db_test = tf.data.Dataset.from_tensor_slices((x_test, y_test)) db_test = db_test.map(process).batch(batch_size) # 直接调用模型 模型的层数已经在自定义里面设置了 net_work = MyModel() net_work.compile( optimizer=optimizers.Adam(lr=0.01), # 设置优化器,并设置学习速率 loss=tf.losses.CategoricalCrossentropy(from_logits=True), # 定义损失函数 metrics=["accuracy"] # 设置模型评估参数 ) # 进行训练 net_work.fit(db, epochs=5, validation_data=db_test, validation_freq=2) # 在测试集上进行验证 net_work.evaluate(db_test) 全部代码 import tensorflow as tf from tensorflow.keras import optimizers, datasets, Sequential, metrics, layers from tensorflow import keras class MyDense(layers.Layer): def __init__(self, inp_dim, outp_dim): super(MyDense, self).__init__() self.kernel = self.add_variable('w', [inp_dim, outp_dim]) self.bias = self.add_variable('b', [outp_dim]) def call(self, inputs, **kwargs): out = inputs @ self.kernel + self.bias return out class MyModel(keras.Model): def __init__(self): super(MyModel, self).__init__() self.fc1 = MyDense(28 * 28, 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, mask=None): x = self.fc1(inputs) x = tf.nn.relu(x) x = self.fc2(x) x = tf.nn.relu(x) x = self.fc3(x) x = tf.nn.relu(x) x = self.fc4(x) x = tf.nn.relu(x) x = self.fc5(x) return x def process(x, y): x = tf.cast(x, dtype=tf.float32) / 255. y = tf.cast(y, dtype=tf.int64) x = tf.reshape(x, [28 * 28]) y = tf.one_hot(y, depth=10) return x, y # 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 batch_size = 128 (x, y), (x_test, y_test) = datasets.mnist.load_data() db = tf.data.Dataset.from_tensor_slices((x, y)) db = db.map(process).shuffle(1000).batch(batch_size) db_test = tf.data.Dataset.from_tensor_slices((x_test, y_test)) db_test = db_test.map(process).batch(batch_size) print(x.shape, y.shape) net_work = MyModel() net_work.compile( optimizer=optimizers.Adam(lr=0.01), loss=tf.losses.CategoricalCrossentropy(from_logits=True), # loss=tf.losses.MSE, metrics=["accuracy"] ) net_work.fit(db, epochs=5, validation_data=db_test, validation_freq=2) net_work.evaluate(db_test)