tree ab2dcf67839080155d1945543ffb20293f84e6e1
parent 4e8098235cb20f7585ff345dea7d7cb6f075d030
author Reed Wanderman-Milne <reedwm@google.com> 1586898700 -0700
committer TensorFlower Gardener <gardener@tensorflow.org> 1586898870 -0700

Change LossScaleOptimizer checkpoint format.

Now the format is identical to as if a LossScaleOptimzier is not used, except that the loss scale is saved with a LossScaleOptimizer. This allows saving checkpoints with a LossScaleOptimizer and restoring without a LossScaleOptimizer, and vice versa.

Checkpoints with LossScaleOptimizers created in older versions of TensorFlow can still be loaded. New checkpoints saved will use the new format.

PiperOrigin-RevId: 306511555
Change-Id: Ie316ab8c4fbfec7babd6f7803d337799d0ff10a5