Details
-
New Feature
-
Status: Closed
-
Minor
-
Resolution: Fixed
-
None
-
None
Description
http://ruder.io/optimizing-gradient-descent/index.html#rmsprop
Learning rate of Adagrad monotonically decreases. Better to support RMSprop as well.
Attachments
Issue Links
- is related to
-
HIVEMALL-101 Separate optimizer implementation
- Closed