Skip to content

Collection of ideas for new features #1

@AndreWeiner

Description

@AndreWeiner

Improvements currently in the pipeline:

  • use functionObject control mechanism in BC
  • create abstraction for DRL components common to all environments (C++)
  • option to save entire simulations
  • implement batch training for PPO agent
  • check and logging of empty trajectories in buffer class
  • timeout option for buffers
  • somehow handle slurm output files
  • read training config from file; allow runtime modifications
  • include default normalization for policy/value inputs
  • allow starting from randomly selected initial times
  • allow learning from multiple different environments

If you have any more suggestions, please feel free to comment on this issue.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions