Skip to content

Latest commit

 

History

History
9 lines (5 loc) · 498 Bytes

TODO.md

File metadata and controls

9 lines (5 loc) · 498 Bytes
Raw

  • Multi-scaling currently does not support propagating additional parameters (for example from smoothers).

  • For the learned smoother multi-scale support would require supporting storing and instantiating different smoothers at the different multi-scale levels. (i.e., level-specific learned deep networks)

  • Write tests for the adaptive smoothers. Test them more and check that the CUDA version in fact works.

  • Fix absolute path in data_manager.py (support data location via a variable).