tf.keras“历史未定义”......但在我看来它已被定义

Question

我正在 TensorFlow 中为我正在学习的入门课程构建一个简单的模型。我找不到错误“历史未定义”的原因。在确定此错误的根源并修复它方面，我将非常感谢任何帮助。

代码如下：

def train_model(model, scaled_train_images, train_labels):
    history = model.fit(scaled_train_images, train_labels, epochs=5, batch_size=256)

# Run function to train the model
train_model(model, scaled_train_images, train_labels)

输出：

Train on 60000 samples

Epoch 1/5
60000/60000 [===] - 56s 936us/sample - loss: 0.0623 - accuracy: 0.9809

Epoch 2/5
60000/60000 [===] - 56s 932us/sample - loss: 0.0538 - accuracy: 0.9838

Epoch 3/5
60000/60000 [===] - 56s 925us/sample - loss: 0.0475 - accuracy: 0.9858

Epoch 4/5
60000/60000 [===] - 55s 923us/sample - loss: 0.0422 - accuracy: 0.9869- los

Epoch 5/5
60000/60000 [===] - 55s 917us/sample - loss: 0.0380 - accuracy: 0.9883

#Load the model history into a pandas DataFrame

frame = pd.DataFrame(history.history)

---------------------------------------------------------------------------
NameError
Traceback (most recent call last)
<ipython-input-22-895ca3f31ddf> in <module>
    1 # Run this cell to load the model history into a pandas DataFrame
    2 
----> 3 frame = pd.DataFrame(history.history)

NameError: name 'history' is not defined

Answer 1

历史变量仅在 train_model 函数内部定义，因此无法在外部访问。

要解决这个问题，请返回：

def train_model(model, scaled_train_images, train_labels):
    return model.fit(scaled_train_images, train_labels, epochs=5, batch_size=256)

history = train_model(model, scaled_train_images, train_labels)

frame = pd.DataFrame(history.history)