Train with custom model in IsaacLab

[PabloVD]

I'm training PPO with skrl in IsaacLab, using this script. All the tasks use by default a yaml config file for the models and agent, which are instantiated through the Runner class. However, I would like to define a custom model in torch (for instance a resnet) to have more flexibility instead of relying on the yaml config. Which is the best way to define that, while keeping the Runner usage?

[Toni-SM]

Hi @PabloVD

You can rely on skrl's standalone scripts for Isaac Lab in order to define your custom model.

https://skrl.readthedocs.io/en/latest/intro/examples.html#nvidia-isaac-lab