Keras model persistence (save, load)

h5py must be installed first:

pip3 install h5py

 

save model

# 假设 model 是训练完毕的模型
# 将模型保存到文件 my_model.h5
model.save('my_model.h5')

 

load model

# 要载入模型，需要导入此函数
from keras.models import load_model

# 从文件 my_model.h5 中载入模型
model = load_model('my_model.h5')

 

Generally speaking, we should train the model in the training environment, and then take the trained model (ie, a file) to the production environment to deploy.