lib/torch/optim/adagrad.rb in torch-rb-0.4.2 vs lib/torch/optim/adagrad.rb in torch-rb-0.5.0

- old
+ new

@@ -55,12 +55,12 @@ clr = group[:lr] / (1 + (state[:step] - 1) * group[:lr_decay]) if grad.sparse? raise NotImplementedYet else - state[:sum].addcmul!(1, grad, grad) + state[:sum].addcmul!(grad, grad, value: 1) std = state[:sum].sqrt.add!(group[:eps]) - p.data.addcdiv!(-clr, grad, std) + p.data.addcdiv!(grad, std, value: -clr) end end end loss