GitLab

* Remove some fp16 examples that don't converge

Default static loss scale of 1.0 (default value) for resnet50 doesn't converge. Either remove example or put static loss scale 128 on it, which is known to converge well.

* Update README.md

* Add reducer class in parallel/distributed.

* Separate DDP and Reducer examples.

* Don't confuse DDP and reducer in example.

Fixes to validation in imagenet example scripts. Precision and loss reporting modified to be consistent with train.