garage.torch.GaussianMLPBaseModule._init_std has wrong dimensionality

rlworkgroup / garage

A toolkit for reproducible reinforcement learning research.

MIT License

1.84k stars 309 forks source link

Closed krzentner closed 3 years ago

krzentner commented 3 years ago

Each dimension of the action space needs to be able to have its variance tuned seperately.

Specifically, this line should broadcast instead of broadcasting in forward

krzentner commented 3 years ago

This was fixed in #2266