search

weidler / RLaSpa

Reinforcement Learning in Latent Space
MIT License
5 stars 1 forks source link

Create different tasks/obstacle maps to "transfer" #17

Closed HansBambel closed 5 years ago

HansBambel commented 5 years ago

By now we only train on one obstacle map. This will in the long run result in overfitting. That's why we should create different maps that can be created when env.reset() is called.

HansBambel commented 5 years ago

Another thing that may speed up learning: reduce the map size.

HansBambel commented 5 years ago

I think it is better to first resolve #20 before working on this.

weidler commented 5 years ago

This is included in the current multi-task pull request