Keras的核心原則是逐步揭示復(fù)雜性，可以在保持相應(yīng)的高級(jí)便利性的同時(shí)，對(duì)操作細(xì)節(jié)進(jìn)行更多控制。當(dāng)我們要自定義fit中的訓(xùn)練算法時(shí)，可以重寫(xiě)模型中的train_step方法，然后調(diào)用fit來(lái)訓(xùn)練模型。
這里以tensorflow2官網(wǎng)中的例子來(lái)說(shuō)明：
import numpy as np
import tensorflow as tf
from tensorflow import keras
x = 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 CustomModel
inputs = 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 usual
model.fit(x, y, epochs=1, shuffle=False)
32/32 [==============================] - 0s 1ms/step - loss: 0.2783 - mae: 0.4257
tensorflow.python.keras.callbacks.History at 0x7ff7edf6dfd0>
這里的loss是tensorflow庫(kù)中實(shí)現(xiàn)了的損失函數(shù)，如果想自定義損失函數(shù)，然后將損失函數(shù)傳入model.compile中，能正常按我們預(yù)想的work嗎？
答案竟然是否定的，而且沒(méi)有錯(cuò)誤提示，只是loss計(jì)算不會(huì)符合我們的預(yù)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.Tensor: shape=(), dtype=float32, numpy=0.11333332>
tf.losses.MSE(a_true, a_pred)
tf.Tensor: shape=(), dtype=float32, numpy=0.11333332>
以上結(jié)果證實(shí)了我們自定義loss的正確性，下面我們直接將自定義的loss置入compile中的loss參數(shù)中，看看會(huì)發(fā)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
tensorflow.python.keras.callbacks.History at 0x7ff7edeb7810>
我們看到，這里的loss與我們與標(biāo)準(zhǔn)的tf.losses.MSE明顯不同。這說(shuō)明我們自定義的loss以這種方式直接傳遞進(jìn)model.compile中，是完全錯(cuò)誤的操作。
正確運(yùn)用自定義loss的姿勢(shì)是什么呢？下面揭曉。
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 CustomModel
inputs = 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 usual
my_model_beta.fit(x, y, epochs=1, shuffle=False)
32/32 [==============================] - 0s 960us/step - loss: 0.2783 - mae: 0.4257
tensorflow.python.keras.callbacks.History at 0x7ff7eda3d810>
終于，通過(guò)跳過(guò)在 compile() 中傳遞損失函數(shù)，而在 train_step 中手動(dòng)完成所有計(jì)算內(nèi)容，我們獲得了與之前默認(rèn)tf.losses.MSE完全一致的輸出，這才是我們想要的結(jié)果。
總結(jié)一下，當(dāng)我們?cè)谀Ｐ椭邢胗米远x的損失函數(shù)，不能直接傳入fit函數(shù)，而是需要在train_step中手動(dòng)傳入，完成計(jì)算過(guò)程。
到此這篇關(guān)于tensorflow2 自定義損失函數(shù)使用的隱藏坑的文章就介紹到這了,更多相關(guān)tensorflow2 自定義損失函數(shù)內(nèi)容請(qǐng)搜索腳本之家以前的文章或繼續(xù)瀏覽下面的相關(guān)文章希望大家以后多多支持腳本之家！
您可能感興趣的文章:- 一小時(shí)學(xué)會(huì)TensorFlow2之基本操作1實(shí)例代碼
- 詳解TensorFlow2實(shí)現(xiàn)前向傳播
- 詳解TensorFlow2實(shí)現(xiàn)線性回歸
- tensorflow2.0實(shí)現(xiàn)復(fù)雜神經(jīng)網(wǎng)絡(luò)(多輸入多輸出nn，Resnet)
- tensorflow2.0教程之Keras快速入門(mén)
- 一小時(shí)學(xué)會(huì)TensorFlow2基本操作之合并分割與統(tǒng)計(jì)
