使用tensorflow2自定义损失函数需要注意什么

Keras的核心原则是逐步揭示复杂性，可以在保持相应的高级便利性的同时，对操作细节进行更多控制。当我们要自定义fit中的训练算法时，可以重写模型中的train_step方法，然后调用fit来训练模型。

这里以tensorflow2官网中的例子来说明：

import numpy as npimport tensorflow as tffrom tensorflow import kerasx = np.random.random((1000, 32))y = np.random.random((1000, 1))class CustomModel(keras.Model):   tf.random.set_seed(100)   def train_step(self, data):       # Unpack the data. Its structure depends on your model and       # on what you pass to `fit()`.       x, y = data       with tf.GradientTape() as tape:           y_pred = self(x, training=True)  # Forward pass           # Compute the loss value           # (the loss function is configured in `compile()`)           loss = self.compiled_loss(y, y_pred, regularization_losses=self.losses)       # Compute gradients       trainable_vars = self.trainable_variables       gradients = tape.gradient(loss, trainable_vars)       # Update weights       self.optimizer.apply_gradients(zip(gradients, trainable_vars))       # Update metrics (includes the metric that tracks the loss)       self.compiled_metrics.update_state(y, y_pred)       # Return a dict mapping metric names to current value       return {m.name: m.result() for m in self.metrics}   # Construct and compile an instance of CustomModelinputs = keras.Input(shape=(32,))outputs = keras.layers.Dense(1)(inputs)model = CustomModel(inputs, outputs)model.compile(optimizer="adam", loss=tf.losses.MSE, metrics=["mae"])# Just use `fit` as usualmodel.fit(x, y, epochs=1, shuffle=False)32/32 [==============================] - 0s 1ms/step - loss: 0.2783 - mae: 0.4257

这里的loss是tensorflow库中实现了的损失函数，如果想自定义损失函数，然后将损失函数传入model.compile中，能正常按我们预想的work吗？

答案竟然是否定的，而且没有错误提示，只是loss计算不会符合我们的预期。

def custom_mse(y_true, y_pred):   return tf.reduce_mean((y_true - y_pred)**2, axis=-1)a_true = tf.constant([1., 1.5, 1.2])a_pred = tf.constant([1., 2, 1.5])custom_mse(a_true, a_pred)tf.losses.MSE(a_true, a_pred)

以上结果证实了我们自定义loss的正确性，下面我们直接将自定义的loss置入compile中的loss参数中，看看会发生什么。

my_model = CustomModel(inputs, outputs)my_model.compile(optimizer="adam", loss=custom_mse, metrics=["mae"])my_model.fit(x, y, epochs=1, shuffle=False)32/32 [==============================] - 0s 820us/step - loss: 0.1628 - mae: 0.3257

我们看到，这里的loss与我们与标准的tf.losses.MSE明显不同。这说明我们自定义的loss以这种方式直接传递进model.compile中，是完全错误的操作。

正确运用自定义loss的姿势是什么呢？下面揭晓。

loss_tracker = keras.metrics.Mean(name="loss")mae_metric = keras.metrics.MeanAbsoluteError(name="mae")class MyCustomModel(keras.Model):   tf.random.set_seed(100)   def train_step(self, data):       # Unpack the data. Its structure depends on your model and       # on what you pass to `fit()`.       x, y = data       with tf.GradientTape() as tape:           y_pred = self(x, training=True)  # Forward pass           # Compute the loss value           # (the loss function is configured in `compile()`)           loss = custom_mse(y, y_pred)           # loss += self.losses       # Compute gradients       trainable_vars = self.trainable_variables       gradients = tape.gradient(loss, trainable_vars)       # Update weights       self.optimizer.apply_gradients(zip(gradients, trainable_vars))              # Compute our own metrics       loss_tracker.update_state(loss)       mae_metric.update_state(y, y_pred)       return {"loss": loss_tracker.result(), "mae": mae_metric.result()}      @property   def metrics(self):       # We list our `Metric` objects here so that `reset_states()` can be       # called automatically at the start of each epoch       # or at the start of `evaluate()`.       # If you don't implement this property, you have to call       # `reset_states()` yourself at the time of your choosing.       return [loss_tracker, mae_metric]   # Construct and compile an instance of CustomModelinputs = keras.Input(shape=(32,))outputs = keras.layers.Dense(1)(inputs)my_model_beta = MyCustomModel(inputs, outputs)my_model_beta.compile(optimizer="adam")# Just use `fit` as usualmy_model_beta.fit(x, y, epochs=1, shuffle=False)32/32 [==============================] - 0s 960us/step - loss: 0.2783 - mae: 0.4257

终于，通过跳过在 compile() 中传递损失函数，而在 train_step 中手动完成所有计算内容，我们获得了与之前默认tf.losses.MSE完全一致的输出，这才是我们想要的结果。