search

weidler / RLaSpa

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

Make pathing task use numpy arrays for faster conversions #9

Closed HansBambel closed 5 years ago

HansBambel commented 5 years ago

By doing #4 I realized that the representation is in Lists. It may be faster to have it in numpy arrays instead. I don't know how hard or if hard at all, but could you maybe look into it?

HansBambel commented 5 years ago

I think a speedup can be achieved if we only update the pixel based representation instead of constructing it every time we need it. Maybe with a flag we either give back the state as coordinates or pixels.

weidler commented 5 years ago

Agreed, that’s what I meant when I said it’s very inefficient atm 🙈 I’ll work on it tomorrow

weidler commented 5 years ago

Done.