araffin / rl-baselines-zoo

A collection of 100+ pre-trained RL agents using Stable Baselines, training and hyperparameter optimization included.
https://stable-baselines.readthedocs.io/
MIT License
1.13k stars 208 forks source link

[feature request] Hyperparameter optimization for pretrained agent #114

Closed Jonathan2021 closed 3 years ago

Jonathan2021 commented 3 years ago

Enabling hyperparameter optimization with the pretrained flag would be a good idea. In my concrete application, I trained an agent on a simplified task and want to continue its training on the real task. For this part of training, running a hyperparameter optimization search would be nice. Of course some parameters, such as the network architecture can't be changed.

EDIT: Sorry this is not the correct version of the zoo. I'll open the issue on SB3 zoo instead.