# How does L1 Loss work in lightGBM

From the paper, lightGBM does a subsampling according to sorted $$|g_i|$$, where $$g_i$$ is the gradient (for the loss function) at a data instance.

My question is that, when the objective is L1 loss/regularization, $$loss_i = |y_i - \hat{y_i}|$$, hence $$g_i$$ is either 1 or -1 (assume no zero loss). Therefore the absolute value of gradient is 1 for any data instance. How to sort then and select instances for the subsample? Or does lightGBM skip the subsampling process if L1 regularization is selected?