TensorFlow saves variables in binary checkpoint files that map variable
names to tensor values. vai_p_tensorflow takes a
checkpoint file as input to load trained weights. The tf.train.Saver
provides methods to specify paths for the checkpoint files
to write to or read from.
Code snippet to call the tf.train.Saver.save
method to save variables to checkpoint files:
with tf.Session() as sess:
# your graph building codes here
# ……
sess.run(train_op)
# Save the variables to disk.
save_path = saver.save(sess, "/tmp/model.ckpt")
print("Model saved in path: %s" % save_path)
The saved checkpoint files look like this:
model.ckpt.data-00000-of-00001
model.ckpt.index
model.ckpt.meta