The high-level API in TensorFlow mainly refers to tensorflow.keras.models
.
This chapter introduces the following details related to tensorflow.keras.models
:
-
Model constructing (using
Sequential
, functional API, or the child class ofModel
) -
Model training (using pre-defined
fit
,train_on_batch
methods, customized training loops, single GPU training, multiple GPU training, and TPU training) -
Model deployment (using tensorflow serving for deployment, using spark (scala) for using tensorflow models)