In line 231 of NeuralLua.lua, the comment says this line is applying bias:
-- aplying BIAS
b = #neural.bias
i = 1
k = #neural.weights
while i <= #neural.weights do
neural.weights[i] = matrix.sum(neural.weights[i],deltas[k])
i = i+1
k = k-1
end
However, it appears to be a cut and paste of the line above, without changing neural.weights[i] to neural.bias[i].
It's not clear how you're applying bias, but it's likely that it can be applied in the same loop that the weights are being applied, instead of requiring another loop.
In line 231 of NeuralLua.lua, the comment says this line is applying bias:
However, it appears to be a cut and paste of the line above, without changing neural.weights[i] to neural.bias[i].
It's not clear how you're applying bias, but it's likely that it can be applied in the same loop that the weights are being applied, instead of requiring another loop.