Darwinbots Forum
Code center => Bugs and fixes => Solved Bugs => Topic started by: Shadowgod2 on November 28, 2013, 02:22:49 AM
-
ok i long handed it and it seems to fix it
-
this is the save file for the generational epigenetic mem copying issue
-
According to what I have so far, when Shadowgods NNet robot reproduces G2 it fails to copy the 15 extra sysvars trough birthtie.
Shadowgod, are you sure there is a birth tie on G2 ?
-
positive i watched it myself and the bot has no way to create any other tie. the test bot you gave me works fine if you don't have the NNet in the same sim and if you do you have to restart to get it to work with only that bot in the sim
-
for some reason it can't get passed into gen 1 anymore.
-
for some reason it can't get passed into gen 1 anymore.
So it stopped working completely now? Can I please get the sim file? This will give me something to work with...
-
yea sure only the first 5 pass down
-
ok back to gen 2 again i reinstalled db and removed all other db copies.
-
ok, taking a look right now...
-
Indeed, forced you robot to reproduce at 5K energy, epi-genetic memory was not copied.
-
How strange, at times the birth tie does not point back to the parent and at times it does.
-
really did not want to do this, seems like I have no choice.
The memory will be stored at birth and exposed if the birth tie exists.
-
so i went back to the original small NNet and tried to fix it and found the problem and when i ran it it got to gen 2 with random mems 0. usually the last 3 input locs were missing but the rest of it (weightings) were there. very strange :huh: