aoaoaoaoa

why all my losses in experiment are NaN? I used the same code, parameters and dataset. like this "iter = 19998/20000, loss_seg = nan, loss_adv_p = nan, loss_D = nan,...