instadeepai / Mava

🦁 A research-friendly codebase for fast experimentation of multi-agent reinforcement learning in JAX
Apache License 2.0
737 stars 90 forks source link

[FEATURE] Set time limit per scenario #1060

Closed SimonDuToit closed 8 months ago

SimonDuToit commented 8 months ago

Currently the time limit, along with other kwargs, are set in the environment config and cannot be specified for individual scenarios. This is a problem as for some environments (e.g. cleaner) a larger scenario may require significantly more timesteps to solve. There should be a way to specify the time limit in individual scenarios. The simplest way to change this is simply moving kwargs to the scenario config and repeating arguments that are the same across scenarios, but we could also treat time limit separately to other kwargs if we know that every environment has it.

sash-a commented 8 months ago

Fixed by #1068