[tf1] save and load parameters

tf.keras

Reference resources https://github.com/tensorflow/docs/blob/529ba4346b8fc5e830e762a2f0ee87b3c345c0c9/site/en/r1/guide/keras.ipynb

# Save weights to a TensorFlow Checkpoint file
model.save_weights('./weights/my_model')

# Restore the model's state,
# this requires a model with the same architecture.
model.load_weights('./weights/my_model')

It can also be saved as Keras HDF5 Format

# Save weights to a HDF5 file
model.save_weights('my_model.h5', save_format='h5')

# Restore the model's state
model.load_weights('my_model.h5')

Be careful ,h5 You may encounter this problem https://stackoverflow.com/questions/53740577/does-any-one-got-attributeerror-str-object-has-no-attribute-decode-whi

namely model.load_weights('my_model.h5') When an error 'str' object has no attribute 'decode'. terms of settlement ：

pip freeze | grep h5py
pip install h5py==2.10.0

tf.train.Saver

You can save the specified parameters . Reference resources https://www.tensorflow.org/api_docs/python/tf/compat/v1/train/Saver

self._saver = tf.train.Saver(var_list=self._get_var_list(), \
				max_to_keep=self.max_tf_checkpoints_to_keep)
self._saver.save(
	self._sess,
	os.path.join(checkpoint_dir, 'tf_ckpt'),
	global_step=self.iteration)
self._saver.restore(self._sess,
                    os.path.join(checkpoint_dir,
                                 'tf_ckpt-{}'.format(iteration_number)))

tf.train.Checkpoint

Don't understand, . Reference resources https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/train/Checkpoint