tensorflow - Keras 中的 KeyError : 'val_acc' on printing history. history.keys()

Question

model.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.Adam(), metrics=['accuracy'])

history = model.fit_generator(train_generator, batch_size, epochs=epochs)

print(history.history.keys())

这导致:['acc', 'loss']

accuracy = history.history['acc']
val_accuracy = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(len(accuracy))

这会产生错误:KeyError: 'val_acc'

为什么我在 history.history.keys() 中看不到 val_acc 和 val_loss？

Answer 1

您没有向 model.fit() 提供任何验证数据，因此没有用于计算 val_acc 的验证数据。您需要将验证数据添加到训练循环中:

history = model.fit_generator(train_generator,
                              batch_size, 
                              epochs,
                              validation_data=validation_generator)