Open thegamecat opened 6 years ago
If i run a deepqlearn regression 1000 times using brain.value_net.to/fromJSON I would expect the actions to improve.
I don't see any significant improvement which is leading me to believe that I am not using the store and re-use correctly.
Does anyone have this working?
If i run a deepqlearn regression 1000 times using brain.value_net.to/fromJSON I would expect the actions to improve.
I don't see any significant improvement which is leading me to believe that I am not using the store and re-use correctly.
Does anyone have this working?