pLogicNet How to understand the implementation of train

How to understand the implementation of train_epoch function?

Open meloncolie opened this issue 4 years ago • 0 comments

trafficstars

First of all, i want to make sure whether this repo is the official implementation of the paper?

According to the implementation of train_epoch:

$logit(h,r,t) = \sigmoid{\frac{\sum_{(h',r',t') \in MB(h,r,t)}w^t(h',r',t')}{N_{MB(h,r,t)}}}$ $w^{t+1} = w^t + \Delta logit(h,r,t)$

I am really confused about how these formulas have been derived according to the paper. I look forward to your reply. Thanks

Feb 26 '21 02:02 meloncolie