原文:https://tensorflow.google.cn/tutorials/keras/save_and_load
Note: 我们的 TensorFlow 社区翻译了这些文档。因为社区翻译是尽力而为, 所以无法保证它们是最准确的,并且反映了最新的 官方英文文档。如果您有改进此翻译的建议, 请提交 pull request 到 tensorflow/docs GitHub 仓库。要志愿地撰写或者审核译文,请加入 [email protected] Google Group。
模型可以在训练期间和训练完成后进行保存。这意味着模型可以从任意中断中恢复,并避免耗费比较长的时间在训练上。保存也意味着您可以共享您的模型,而其他人可以通过您的模型来重新创建工作。在发布研究模型和技术时,大多数机器学习从业者分享:
- 用于创建模型的代码
- 模型训练的权重 (weight) 和参数 (parameters) 。
共享数据有助于其他人了解模型的工作原理,并使用新数据自行尝试。
注意:小心不受信任的代码——Tensorflow 模型是代码。有关详细信息,请参阅 安全使用 Tensorflow。
保存 Tensorflow 的模型有许多方法——具体取决于您使用的 API。本指南使用 tf.keras, 一个高级 API 用于在 Tensorflow 中构建和训练模型。有关其他方法的实现,请参阅 TensorFlow 保存和恢复指南或保存到 eager。
安装并导入 Tensorflow 和依赖项:
pip install -q pyyaml h5py # 以 HDF5 格式保存模型所必须
WARNING: You are using pip version 20.2.2; however, version 20.2.3 is available.
You should consider upgrading via the '/tmpfs/src/tf_docs_env/bin/python -m pip install --upgrade pip' command.
import os
import tensorflow as tf
from tensorflow import keras
print(tf.version.VERSION)
2.3.0
要演示如何保存和加载权重,您将使用 MNIST 数据集. 要加快运行速度,请使用前 1000 个示例:
(train_images, train_labels), (test_images, test_labels) = tf.keras.datasets.mnist.load_data()
train_labels = train_labels[:1000]
test_labels = test_labels[:1000]
train_images = train_images[:1000].reshape(-1, 28 * 28) / 255.0
test_images = test_images[:1000].reshape(-1, 28 * 28) / 255.0
首先构建一个简单的序列(sequential)模型:
# 定义一个简单的序列模型
def create_model():
model = tf.keras.models.Sequential([
keras.layers.Dense(512, activation='relu', input_shape=(784,)),
keras.layers.Dropout(0.2),
keras.layers.Dense(10)
])
model.compile(optimizer='adam',
loss=tf.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
return model
# 创建一个基本的模型实例
model = create_model()
# 显示模型的结构
model.summary()
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense (Dense) (None, 512) 401920
_________________________________________________________________
dropout (Dropout) (None, 512) 0
_________________________________________________________________
dense_1 (Dense) (None, 10) 5130
=================================================================
Total params: 407,050
Trainable params: 407,050
Non-trainable params: 0
_________________________________________________________________
您可以使用训练好的模型而无需从头开始重新训练,或在您打断的地方开始训练,以防止训练过程没有保存。 tf.keras.callbacks.ModelCheckpoint
允许在训练的过程中和结束时回调保存的模型。
创建一个只在训练期间保存权重的 tf.keras.callbacks.ModelCheckpoint
回调:
checkpoint_path = "training_1/cp.ckpt"
checkpoint_dir = os.path.dirname(checkpoint_path)
# 创建一个保存模型权重的回调
cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,
save_weights_only=True,
verbose=1)
# 使用新的回调训练模型
model.fit(train_images,
train_labels,
epochs=10,
validation_data=(test_images,test_labels),
callbacks=[cp_callback]) # 通过回调训练
# 这可能会生成与保存优化程序状态相关的警告。
# 这些警告(以及整个笔记本中的类似警告)
# 是防止过时使用,可以忽略。
Epoch 1/10
29/32 [==========================>...] - ETA: 0s - loss: 1.1844 - accuracy: 0.6595
Epoch 00001: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 8ms/step - loss: 1.1300 - accuracy: 0.6770 - val_loss: 0.7189 - val_accuracy: 0.7780
Epoch 2/10
30/32 [===========================>..] - ETA: 0s - loss: 0.4232 - accuracy: 0.8792
Epoch 00002: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 5ms/step - loss: 0.4216 - accuracy: 0.8800 - val_loss: 0.5160 - val_accuracy: 0.8470
Epoch 3/10
29/32 [==========================>...] - ETA: 0s - loss: 0.2964 - accuracy: 0.9149
Epoch 00003: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.2988 - accuracy: 0.9170 - val_loss: 0.4753 - val_accuracy: 0.8560
Epoch 4/10
29/32 [==========================>...] - ETA: 0s - loss: 0.2057 - accuracy: 0.9494
Epoch 00004: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.2086 - accuracy: 0.9500 - val_loss: 0.4375 - val_accuracy: 0.8600
Epoch 5/10
29/32 [==========================>...] - ETA: 0s - loss: 0.1512 - accuracy: 0.9666
Epoch 00005: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.1488 - accuracy: 0.9680 - val_loss: 0.4275 - val_accuracy: 0.8660
Epoch 6/10
30/32 [===========================>..] - ETA: 0s - loss: 0.1130 - accuracy: 0.9823
Epoch 00006: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.1134 - accuracy: 0.9820 - val_loss: 0.4309 - val_accuracy: 0.8630
Epoch 7/10
29/32 [==========================>...] - ETA: 0s - loss: 0.0829 - accuracy: 0.9925
Epoch 00007: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.0838 - accuracy: 0.9920 - val_loss: 0.4079 - val_accuracy: 0.8680
Epoch 8/10
29/32 [==========================>...] - ETA: 0s - loss: 0.0624 - accuracy: 0.9946
Epoch 00008: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.0627 - accuracy: 0.9950 - val_loss: 0.4176 - val_accuracy: 0.8690
Epoch 9/10
29/32 [==========================>...] - ETA: 0s - loss: 0.0520 - accuracy: 0.9946
Epoch 00009: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.0508 - accuracy: 0.9950 - val_loss: 0.4600 - val_accuracy: 0.8450
Epoch 10/10
29/32 [==========================>...] - ETA: 0s - loss: 0.0462 - accuracy: 0.9968
Epoch 00010: saving model to training_1/cp.ckpt
32/32 [==============================] - 0s 4ms/step - loss: 0.0459 - accuracy: 0.9970 - val_loss: 0.4378 - val_accuracy: 0.8660
<tensorflow.python.keras.callbacks.History at 0x7fe7b286b710>
这将创建一个 TensorFlow checkpoint 文件集合,这些文件在每个 epoch 结束时更新:
ls {checkpoint_dir}
checkpoint cp.ckpt.data-00000-of-00001 cp.ckpt.index
创建一个新的未经训练的模型。仅恢复模型的权重时,必须具有与原始模型具有相同网络结构的模型。由于模型具有相同的结构,您可以共享权重,尽管它是模型的不同实例。 现在重建一个新的未经训练的模型,并在测试集上进行评估。未经训练的模型将在机会水平(chance levels)上执行(准确度约为 10%):
# 创建一个基本模型实例
model = create_model()
# 评估模型
loss, acc = model.evaluate(test_images, test_labels, verbose=2)
print("Untrained model, accuracy: {:5.2f}%".format(100*acc))
32/32 - 0s - loss: 2.3734 - accuracy: 0.0990
Untrained model, accuracy: 9.90%
然后从 checkpoint 加载权重并重新评估:
# 加载权重
model.load_weights(checkpoint_path)
# 重新评估模型
loss,acc = model.evaluate(test_images, test_labels, verbose=2)
print("Restored model, accuracy: {:5.2f}%".format(100*acc))
32/32 - 0s - loss: 0.4378 - accuracy: 0.8660
Restored model, accuracy: 86.60%
回调提供了几个选项,为 checkpoint 提供唯一名称并调整 checkpoint 频率。
训练一个新模型,每五个 epochs 保存一次唯一命名的 checkpoint :
# 在文件名中包含 epoch (使用 `str.format`)
checkpoint_path = "training_2/cp-{epoch:04d}.ckpt"
checkpoint_dir = os.path.dirname(checkpoint_path)
# 创建一个回调,每 5 个 epochs 保存模型的权重
cp_callback = tf.keras.callbacks.ModelCheckpoint(
filepath=checkpoint_path,
verbose=1,
save_weights_only=True,
period=5)
# 创建一个新的模型实例
model = create_model()
# 使用 `checkpoint_path` 格式保存权重
model.save_weights(checkpoint_path.format(epoch=0))
# 使用新的回调训练模型
model.fit(train_images,
train_labels,
epochs=50,
callbacks=[cp_callback],
validation_data=(test_images,test_labels),
verbose=0)
WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
Epoch 00005: saving model to training_2/cp-0005.ckpt
Epoch 00010: saving model to training_2/cp-0010.ckpt
Epoch 00015: saving model to training_2/cp-0015.ckpt
Epoch 00020: saving model to training_2/cp-0020.ckpt
Epoch 00025: saving model to training_2/cp-0025.ckpt
Epoch 00030: saving model to training_2/cp-0030.ckpt
Epoch 00035: saving model to training_2/cp-0035.ckpt
Epoch 00040: saving model to training_2/cp-0040.ckpt
Epoch 00045: saving model to training_2/cp-0045.ckpt
Epoch 00050: saving model to training_2/cp-0050.ckpt
<tensorflow.python.keras.callbacks.History at 0x7fe8021c76a0>
现在查看生成的 checkpoint 并选择最新的 checkpoint :
ls {checkpoint_dir}
checkpoint cp-0025.ckpt.index
cp-0000.ckpt.data-00000-of-00001 cp-0030.ckpt.data-00000-of-00001
cp-0000.ckpt.index cp-0030.ckpt.index
cp-0005.ckpt.data-00000-of-00001 cp-0035.ckpt.data-00000-of-00001
cp-0005.ckpt.index cp-0035.ckpt.index
cp-0010.ckpt.data-00000-of-00001 cp-0040.ckpt.data-00000-of-00001
cp-0010.ckpt.index cp-0040.ckpt.index
cp-0015.ckpt.data-00000-of-00001 cp-0045.ckpt.data-00000-of-00001
cp-0015.ckpt.index cp-0045.ckpt.index
cp-0020.ckpt.data-00000-of-00001 cp-0050.ckpt.data-00000-of-00001
cp-0020.ckpt.index cp-0050.ckpt.index
cp-0025.ckpt.data-00000-of-00001
latest = tf.train.latest_checkpoint(checkpoint_dir)
latest
'training_2/cp-0050.ckpt'
注意: 默认的 tensorflow 格式仅保存最近的 5 个 checkpoint 。
如果要进行测试,请重置模型并加载最新的 checkpoint :
# 创建一个新的模型实例
model = create_model()
# 加载以前保存的权重
model.load_weights(latest)
# 重新评估模型
loss, acc = model.evaluate(test_images, test_labels, verbose=2)
print("Restored model, accuracy: {:5.2f}%".format(100*acc))
32/32 - 0s - loss: 0.4836 - accuracy: 0.8750
Restored model, accuracy: 87.50%
上述代码将权重存储到 checkpoint—— 格式化文件的集合中,这些文件仅包含二进制格式的训练权重。 Checkpoints 包含:
- 一个或多个包含模型权重的分片。
- 索引文件,指示哪些权重存储在哪个分片中。
如果你只在一台机器上训练一个模型,你将有一个带有后缀的碎片: .data-00000-of-00001
您将了解如何将权重加载到模型中。使用 Model.save_weights
方法手动保存它们同样简单。默认情况下, tf.keras
和 save_weights
特别使用 TensorFlow checkpoints 格式 .ckpt
扩展名和 ( 保存在 HDF5 扩展名为 .h5
保存并序列化模型 ):
# 保存权重
model.save_weights('./checkpoints/my_checkpoint')
# 创建模型实例
model = create_model()
# 恢复权重
model.load_weights('./checkpoints/my_checkpoint')
# 评估模型
loss,acc = model.evaluate(test_images, test_labels, verbose=2)
print("Restored model, accuracy: {:5.2f}%".format(100*acc))
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
32/32 - 0s - loss: 0.4836 - accuracy: 0.8750
Restored model, accuracy: 87.50%
调用 model.save
将保存模型的结构,权重和训练配置保存在单个文件/文件夹中。这可以让您导出模型,以便在不访问原始 Python 代码*的情况下使用它。因为优化器状态(optimizer-state)已经恢复,您可以从中断的位置恢复训练。
整个模型可以以两种不同的文件格式(SavedModel
和 HDF5
)进行保存。需要注意的是 TensorFlow 的 SavedModel
格式是 TF2.x. 中的默认文件格式。但是,模型仍可以以 HDF5
格式保存。下面介绍了以两种文件格式保存整个模型的更多详细信息。
保存完整模型会非常有用——您可以在 TensorFlow.js(Saved Model, HDF5)加载它们,然后在 web 浏览器中训练和运行它们,或者使用 TensorFlow Lite 将它们转换为在移动设备上运行(Saved Model, HDF5)
*自定义对象(例如,子类化模型或层)在保存和加载时需要特别注意。请参阅下面的保存自定义对象部分
SavedModel 格式是序列化模型的另一种方法。以这种格式保存的模型,可以使用 tf.keras.models.load_model
还原,并且模型与 TensorFlow Serving 兼容。SavedModel 指南详细介绍了如何提供/检查 SavedModel。以下部分说明了保存和还原模型的步骤。
# 创建并训练一个新的模型实例。
model = create_model()
model.fit(train_images, train_labels, epochs=5)
# 将整个模型另存为 SavedModel。
!mkdir -p saved_model
model.save('saved_model/my_model')
Epoch 1/5
32/32 [==============================] - 0s 2ms/step - loss: 1.1705 - accuracy: 0.6690
Epoch 2/5
32/32 [==============================] - 0s 2ms/step - loss: 0.4326 - accuracy: 0.8780
Epoch 3/5
32/32 [==============================] - 0s 2ms/step - loss: 0.2910 - accuracy: 0.9190
Epoch 4/5
32/32 [==============================] - 0s 2ms/step - loss: 0.2045 - accuracy: 0.9520
Epoch 5/5
32/32 [==============================] - 0s 2ms/step - loss: 0.1538 - accuracy: 0.9650
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow/python/training/tracking/tracking.py:111: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
This property should not be used in TensorFlow 2.0, as updates are applied automatically.
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow/python/training/tracking/tracking.py:111: Layer.updates (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.
Instructions for updating:
This property should not be used in TensorFlow 2.0, as updates are applied automatically.
INFO:tensorflow:Assets written to: saved_model/my_model/assets
SavedModel 格式是一个包含 protobuf 二进制文件和 Tensorflow 检查点(checkpoint)的目录。检查保存的模型目录:
# my_model 文件夹
ls saved_model
# 包含一个 assets 文件夹,saved_model.pb,和变量文件夹。
ls saved_model/my_model
my_model
assets saved_model.pb variables
从保存的模型重新加载新的 Keras 模型:
new_model = tf.keras.models.load_model('saved_model/my_model')
# 检查其架构
new_model.summary()
Model: "sequential_5"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_10 (Dense) (None, 512) 401920
_________________________________________________________________
dropout_5 (Dropout) (None, 512) 0
_________________________________________________________________
dense_11 (Dense) (None, 10) 5130
=================================================================
Total params: 407,050
Trainable params: 407,050
Non-trainable params: 0
_________________________________________________________________
还原的模型使用与原始模型相同的参数进行编译。 尝试使用加载的模型运行评估和预测:
# 评估还原的模型
loss, acc = new_model.evaluate(test_images, test_labels, verbose=2)
print('Restored model, accuracy: {:5.2f}%'.format(100*acc))
print(new_model.predict(test_images).shape)
32/32 - 0s - loss: 0.4630 - accuracy: 0.0890
Restored model, accuracy: 8.90%
(1000, 10)
Keras 使用 HDF5 标准提供了一种基本的保存格式。
# 创建并训练一个新的模型实例
model = create_model()
model.fit(train_images, train_labels, epochs=5)
# 将整个模型保存为 HDF5 文件。
# '.h5' 扩展名指示应将模型保存到 HDF5。
model.save('my_model.h5')
Epoch 1/5
32/32 [==============================] - 0s 2ms/step - loss: 1.1465 - accuracy: 0.6560
Epoch 2/5
32/32 [==============================] - 0s 2ms/step - loss: 0.4152 - accuracy: 0.8850
Epoch 3/5
32/32 [==============================] - 0s 2ms/step - loss: 0.2801 - accuracy: 0.9280
Epoch 4/5
32/32 [==============================] - 0s 2ms/step - loss: 0.2108 - accuracy: 0.9480
Epoch 5/5
32/32 [==============================] - 0s 2ms/step - loss: 0.1520 - accuracy: 0.9660
现在,从该文件重新创建模型:
# 重新创建完全相同的模型,包括其权重和优化程序
new_model = tf.keras.models.load_model('my_model.h5')
# 显示网络结构
new_model.summary()
Model: "sequential_6"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_12 (Dense) (None, 512) 401920
_________________________________________________________________
dropout_6 (Dropout) (None, 512) 0
_________________________________________________________________
dense_13 (Dense) (None, 10) 5130
=================================================================
Total params: 407,050
Trainable params: 407,050
Non-trainable params: 0
_________________________________________________________________
检查其准确率(accuracy):
loss, acc = new_model.evaluate(test_images, test_labels, verbose=2)
print('Restored model, accuracy: {:5.2f}%'.format(100*acc))
32/32 - 0s - loss: 0.4639 - accuracy: 0.0840
Restored model, accuracy: 8.40%
Keras 通过检查网络结构来保存模型。这项技术可以保存一切:
- 权重值
- 模型的架构
- 模型的训练配置(您传递给编译的内容)
- 优化器及其状态(如果有的话)(这使您可以在中断的地方重新开始训练)
Keras 无法保存 v1.x
优化器(来自 tf.compat.v1.train
),因为它们与检查点不兼容。对于 v1.x 优化器,您需要在加载-失去优化器的状态后,重新编译模型。
如果使用的是 SavedModel 格式,则可以跳过此部分。HDF5 和 SavedModel 之间的主要区别在于,HDF5 使用对象配置保存模型结构,而 SavedModel 保存执行图。因此,SavedModel 能够保存自定义对象,例如子类化模型和自定义层,而无需原始代码。
要将自定义对象保存到 HDF5,必须执行以下操作:
- 在对象中定义一个
get_config
方法,以及可选的from_config
类方法。get_config(self)
返回重新创建对象所需的参数的 JSON 可序列化字典。from_config(cls, config)
使用从 get_config 返回的 config 来创建一个新对象。默认情况下,此函数将使用 config 作为初始化 kwargs(return cls(**config)
)。
- 加载模型时,将对象传递给
custom_objects
参数。参数必须是将字符串类名称映射到 Python 类的字典。例如,tf.keras.models.load_model(path, custom_objects={'CustomLayer': CustomLayer})
有关自定义对象和 get_config
的示例,请参见从头开始编写层和模型教程。
# MIT License
#
# Copyright (c) 2017 François Chollet
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense,
# and/or sell copies of the Software, and to permit persons to whom the
# Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
# DEALINGS IN THE SOFTWARE.