nprizeadditions
nprizeadditions copied to clipboard
small bug was found
line 225 in rbm.c should be fixed from visbiases[j][i] = 0.02 * randn() - 0.004;
to visbiases[j][i] = log( ((double)moviercount[j*SOFTMAX+i]) / ((double) mtot) );
?
Hmmm, possibly. Does that appear to work better? I think I found random init working better.
I've realized a while ago that my softmax calc was off given better understanding of intent.
i didn't note the change but it really worked better