feature: accept any structure of results

DeNA / HandyRL

HandyRL is a handy and simple framework based on Python and PyTorch for distributed reinforcement learning that is applicable to your own environments.

MIT License

282 stars 41 forks source link

feature: accept any structure of results #327

Closed YuriCat closed 1 year ago

YuriCat commented 1 year ago

This commit enables us to accept outputs like {'estimate_input': {'a': Tensor, 'b': Tensor}}

YuriCat commented 1 year ago

This PR is not enough.