Hi, they are the same sigmoid function (with the same formula if you checked), but the 1. is in pytorch functional api - you know you can define network models in pytorch in standard sequential way or functional way. (actually these 2 ways are also there in all other frameworks, so not pytorch specific)