quancore / social-lstm

Social LSTM implementation in PyTorch

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

about the train loss

bbidong opened this issue · comments

In the training, i find the loss less than 0 sometimes. i guess the smaller the loss, the better(eg: loss=-10 is better than loss=-5).
Am i right ?? @quancore

In the training, i find the loss less than 0 sometimes. i guess the smaller the loss, the better(eg: loss=-10 is better than loss=-5).
Am i right ?? @quancore

sure ,the reason of loss less than 0 is that -log(p|o,u p)