Closed ICG0401 closed 2 years ago
It’s a generative model, frequent evaluation at some points can yield better results.
On Fri, Apr 22, 2022 at 6:05 AM ICG0401 @.***> wrote:
In the process of training, the value of loss is from positive to negative. However, in this paper, the loss function uses negative log likelihood. Why does the loss value appear to be negative?
— Reply to this email directly, view it on GitHub https://github.com/abduallahmohamed/Social-STGCNN/issues/59, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFIVLGYIFTATXRKNYUQYJE3VGKPY7ANCNFSM5UCITRBA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
--
Abduallah Mohamed abduallahmohamed.com
In the process of training, the value of loss is from positive to negative. However, in this paper, the loss function uses negative log likelihood. Why does the loss value appear to be negative?