Tensorflow2.0 entry and practical study notes (supplement)-model saving and restoration

 

1 Introduction

This method saves all of the following:
1. Weight value

2. Model configuration (architecture)

3. Optimizer configuration
 

1.1 model.save()

Save schema only

Recovery model (reconstruction):

Too bad to save the state of the optimizer

1.2 Save checkpoint

2 Reload the model

Save checkpoints in custom training

checkpoin t where to save changes

Save place:

 

Guess you like

Origin blog.csdn.net/qq_37457202/article/details/107977144