About loss

[EJHyun]

Hi I'm curious about your loss because when i read your paper, the loss's shape is like

• sigma(positive / sigma(negatives)) this

but when i printed your loss in TKG_Module.py, like

def training_step(self, batch_time, batch_idx):

gc.collect()

    loss = self.forward(batch_time)
    print("\n",loss.item())

and i see the loss values are all positive values. is the loss your model using for training exactly same with the paper???

(I'm just a student. I'm just curious about it so please don't feel bad about the question)

[JiapengWu]

Yes they are the same. Maximizing the negative of the loss is equivalent to minimizing the loss.

On Sep 14, 2021, at 10:05 AM, Noc_Up @.**@.>> wrote:

Hi I'm curious about your loss because when i read your paper, the loss's shape is like

• sigma(positive / sigma(negatives)) this

but when i printed your loss in TKG_Module.py, like

def training_step(self, batch_time, batch_idx):

gc.collect()

loss = self.forward(batch_time) print("\n",loss.item())

and i see the loss values are all positive values. is the loss your model using for training exactly same with the paper???

(I'm just a student. I'm just curious about it so please don't feel bad about the question)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/JiapengWu/TeMP/issues/12, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AFN2LV2XKCZ4MY57YZUIN4TUB5JBPANCNFSM5EAHVGVQ. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.