Is there a way to tell TFLearn to save checkpoints in the V2 format? I am using the current (r0.12.1) release of TensorFlow. If you follow a simple example, such as:
https://www.tensorflow.org/tutorials/tflearn/
You will get flooded with:
WARNING:tensorflow:*******************************************************
WARNING:tensorflow:TensorFlow's V1 checkpoint format has been deprecated.
WARNING:tensorflow:Consider switching to the more efficient V2 format:
WARNING:tensorflow: `tf.train.Saver(write_version=tf.train.SaverDef.V2)`
WARNING:tensorflow:now on by default.
WARNING:tensorflow:*******************************************************
While the warnings are easy enough to ignore. I would like to save these files into the correct version so that I do not have to upgrade them later on.
The issue is that some of the higher level API calls are using the older checkpoint format. At this point, in r.0.12.1, it is best to suppress or ignore the warnings.
This might be fixed in the new version https://www.tensorflow.org/versions/r1.0/