ml-agents add an option to change timeout before unity responds

Is your feature request related to a problem? Please describe.

mlagents-learn times out when Unity play is not pressed. Sometimes unity play takes 30 seconds to start, and MLAgents-learn times out by then

Describe the solution you'd like Please add a --timeout or some similar option when starting mlagents-learn

Jan 28 '22 22:01 dmitryponv

Great suggestion. This is also useful as while debugging in Unity editor, the trainer might disconnect on pause

Jan 30 '22 22:01 CubeMD

Thanks for reaching out! We've logged this request internally as MLA-2401.

Feb 09 '22 02:02 miguelalonsojr

please also support turning the timeout off :)

Mar 28 '22 22:03 AVOlight

I guess there's still no update on this, right? Such a simple feature would be extremely useful, especially when starting a large batch of standalone builds

Jul 27 '23 14:07 danielebanovaz

I see it has been now "closed as complete". Does it mean that request MLA-2401 has been fully implemented and it's available in the upcoming release of ML-Agents?

Oct 09 '23 06:10 danielebanovaz

No. Thanks for reaching out. This was closed en masse to clean up old issues and requests. Re-opening as this should be an easy addition. No ETA on implementation though. But will keep open.

Oct 09 '23 11:10 miguelalonsojr

Thank you very much for clarifying!

Oct 09 '23 11:10 danielebanovaz

The update was easy enough. Here's the PR: https://github.com/Unity-Technologies/ml-agents/pull/5991

Oct 09 '23 14:10 miguelalonsojr

This feature has been implemented.

Oct 10 '23 13:10 miguelalonsojr