huawei-noah / SMARTS

Scalable Multi-Agent RL Training School for Autonomous Driving

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[Feature request] Simplified experiment configuration

Gamenot opened this issue · comments

Motivation

It is our desire to provide replicable and easy configuration for smarts examples, environment, and engine configuration.

Proposal

We should attempt to integrate hydra for configuration options for examples.

Alternatives

No response

Additional context

This comes with some concerns:

  • hydra-core has some additional dependencies we do not want in smarts (we should use it only at the experiment level)
  • click and hydra are incompatible at the command line interface level, however we may be able to use [hydra compose](https://hydra.cc/docs/advanced/compose_api/ to combine the two.
  • Directly supplying engine configurations might be useful but would quickly clutter the configuration.