机器学习 tensorflow 2 的model 的保存，导入和继续培训

关于保存和导入可以参看 https://www.tensorflow.org/guide/keras/save_and_serialize

# Save the model
model.save('path_to_my_model.h5')

# Recreate the exact same model purely from the file
new_model = keras.models.load_model('path_to_my_model.h5')

知道了保存和导入，我关心的是再培训。培训次数太多，不知道什么时候结束。培训了很久，基本达到要求了，只要再培训一点点了，就不知道怎么再继续培训。我找呀找，后来找到很简单，就是继续培训 model.fit 就好了。

比如

model.fit(xs, ys, epochs=100)

或者： 

INIT_LR = 0.01
EPOCHS = 15
BS = 32

# initialize the model and optimizer (you'll want to use
# binary_crossentropy for 2-class classification)
print("[INFO] retraining network...")
#opt = SGD(lr=INIT_LR, decay=INIT_LR / EPOCHS)
#model.compile(loss="categorical_crossentropy", optimizer=opt,
#	metrics=["accuracy"])

# train the network
H = model.fit_generator(aug.flow(trainX, trainY, batch_size=BS),
	validation_data=(testX, testY), steps_per_epoch=len(trainX) // BS,
	epochs=EPOCHS)

注意的是不要再compile 等初始化了，否则以前的培训不起作用了。

下面再说说保存和导入：

全部保存在一个文件可以用

model.save('path_to_saved_model',save_format='tf')

导入的方式还是一样，new_model=keras.models.load_model('path_to_saved_model')

示范代码如下：

# Export the model to a SavedModel
model.save('path_to_saved_model', save_format='tf')

# Recreate the exact same model
new_model = keras.models.load_model('path_to_saved_model')

# Check that the state is preserved
new_predictions = new_model.predict(x_test)
np.testing.assert_allclose(predictions, new_predictions, rtol=1e-6, atol=1e-6)

# Note that the optimizer state is preserved as well:
# you can resume training where you left off.

开头的那种方式会保存成一个目录，model.save('path_to_saved_model')，会保存一个目录path_to_saved_model ，然后其下会有2个目录和一个文件save_model.pb。

path_to_saved_model
assets  saved_model.pb  variables

注意：

虽然我们用tensorflow2的也是keras，但是其保存的模型和 直接的keras 保存的模型不通用，有时会报错。

模型保存到JSON

JSON是一种用于分层描述数据的简单文件格式。

Keras提供了使用to_json（）函数使用JSON格式描述任何模型的功能。可以通过model_from_json（）函数保存该函数，该函数将根据JSON规范创建一个新模型。

使用save_weights（）函数直接从模型中保存权重到model.h5，然后使用load_weights（）函数加载权重。

from tensorflow.keras.models import model_from_json

from tensorflow.keras.models import model_from_json
#-----other code------
# serialize model to JSON
print("save to Json")
model_json = model.to_json()
with open("model.json", "w") as json_file:
    json_file.write(model_json)
# serialize weights to HDF5
model.save_weights("model.h5")
print("Saved model to disk")
 
# later...
 
# load json and create model
print('load from json')
json_file = open('model.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
print('load from h5')
loaded_model = model_from_json(loaded_model_json)
# load weights into new model
loaded_model.load_weights("model.h5")
print("Loaded model from disk")

保存模型到 YAML

这个例子和保存到 JSON 一样，只是改为 YAML 格式，要求安装好了 PyYAML 5 

from tensorflow.keras.models import model_from_yaml
#-----other code  -------
# serialize model to YAML
model_yaml = model.to_yaml()
with open("model.yaml", "w") as yaml_file:
    yaml_file.write(model_yaml)
# serialize weights to HDF5
model.save_weights("model.h5")
print("Saved model to disk")
 
# later...
 
# load YAML and create model
print('load from yaml')
yaml_file = open('model.yaml', 'r')
loaded_model_yaml = yaml_file.read()
yaml_file.close()
loaded_model = model_from_yaml(loaded_model_yaml)
# load weights into new model
print('load from h5')
loaded_model.load_weights("model.h5")
print("Loaded model from disk")