> For the complete documentation index, see [llms.txt](https://bing-dong.gitbook.io/tensorflow/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://bing-dong.gitbook.io/tensorflow/tf.keras--callbacks.md).

# tf.keras--callbacks

```python
# 回调函数
# 是在模型的训练过程中进行调用实现的，作为一个参数传入model.fit()中

import matplotlib as mpl
import matplotlib.pyplot as plt
%matplotlib inline
import numpy as np
import sklearn
import pandas as pd
import os
import sys
import time
import tensorflow as tf
from tensorflow import keras
```

```python
# 获取图片数据集
fashion_mnist = keras.datasets.fashion_mnist
(x_train_all,y_train_all),(x_test,y_test) = fashion_mnist.load_data()
# 将训练集拆分为训练集和验证集 共60000张图片，前5000张为验证集，后55000张为训练集
x_valid,x_train = x_train_all[:5000],x_train_all[5000:]
y_valid,y_train = y_train_all[:5000],y_train_all[5000:]
```

```python
from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
x_train_scaled = scaler.fit_transform(x_train.astype(np.float32).reshape(-1,1)).reshape(-1,28,28) 
x_valid_scaled = scaler.transform(x_valid.astype(np.float32).reshape(-1,1)).reshape(-1,28,28)
x_test_scaled = scaler.transform(x_test.astype(np.float32).reshape(-1,1)).reshape(-1,28,28)
```

```python
model = keras.models.Sequential()

model = keras.models.Sequential([
    keras.layers.Flatten(input_shape=[28, 28]),
    keras.layers.Dense(300,activation='relu'),
    keras.layers.Dense(100,activation='relu'),
    keras.layers.Dense(10,activation='softmax')
])

model.compile(loss='sparse_categorical_crossentropy', 
             optimizer = 'adam',
             metrics = ["accuracy"])
```

```python
# 这里使用三个callback函数，TensorBoard, EarlyStopping, ModelCheckpoint
# Tensorboard需要一个文件夹存数据，ModelCheckpoint需要一个文件存数据
# logdir = './callbacks'  此处报错，改用下行
logdir = os.path.join("callbacks")
if not os.path.exists(logdir):
    os.mkdir(logdir)
output_model_file = os.path.join(logdir, "fashion_mnist_model.h5")

callbacks = [
    keras.callbacks.TensorBoard(logdir),
    keras.callbacks.ModelCheckpoint(output_model_file, save_best_only = True), # save_best_only保存最好的模型，否则默认保存最近训练的模型
#     如果连续5次迭代的变化值都小于阈值min_delta，则停止训练
    keras.callbacks.EarlyStopping(monitor='val_accuracy', patience=5, min_delta=1e-2)
]
# 开始训练  callbacks作为参数填入
history = model.fit(x_train_scaled, y_train, epochs=30, validation_data=(x_valid_scaled, y_valid), callbacks = callbacks)
```

```
# 可以看出5次val_accuracy的值都没有提高1e-2，故回调函数EarlyStopping停止了训练
Train on 55000 samples, validate on 5000 samples
Epoch 1/30
55000/55000 [==============================] - 7s 119us/sample - loss: 0.0884 - accuracy: 0.9666 - val_loss: 0.6621 - val_accuracy: 0.8948
Epoch 2/30
55000/55000 [==============================] - 6s 117us/sample - loss: 0.0893 - accuracy: 0.9677 - val_loss: 0.6010 - val_accuracy: 0.9000
Epoch 3/30
55000/55000 [==============================] - 6s 115us/sample - loss: 0.0848 - accuracy: 0.9681 - val_loss: 0.6775 - val_accuracy: 0.8916
Epoch 4/30
55000/55000 [==============================] - 6s 116us/sample - loss: 0.0810 - accuracy: 0.9701 - val_loss: 0.6324 - val_accuracy: 0.8916
Epoch 5/30
55000/55000 [==============================] - 6s 116us/sample - loss: 0.0822 - accuracy: 0.9687 - val_loss: 0.6530 - val_accuracy: 0.8936
Epoch 6/30
55000/55000 [==============================] - 6s 118us/sample - loss: 0.0824 - accuracy: 0.9695 - val_loss: 0.6708 - val_accuracy: 0.8952
```