For ds2, we should not involve padding data when computing loss and gradient #414
Description
To solve the problem, two things should be done.
- Expose SubSequenceLayer (including bug fixes) Bug fix for SubSequnceLayer Paddle#5335
- Add a ScaleSubRegionLayer (support set value for given indices) Add ScaleSubRegion Layer. Paddle#5416
- Modify the network configuration. Support padding removing. #444
- Adapt tune.py Adapt tuning script to padding removing #444 #447