Saving baseline in main.py

tristandeleu / pytorch-maml-rl

Reinforcement Learning with Model-Agnostic Meta-Learning in Pytorch

MIT License

827 stars 158 forks source link

Saving baseline in main.py #17

Closed zhanpenghe closed 6 years ago

zhanpenghe commented 6 years ago

I think to do the k shot adaptation, we need to also save the baseline.

tristandeleu commented 6 years ago

In this specific implementation, the baseline is always trained from scratch so there is no need to save the baseline. But you could definitely meta-learn the baseline as well, and in that case you'd need to save it.