if we have more user features or item features,how to do for input file

[lvjunmei]

information as follows:
userid user_features itemid item_feature rating

[rgeorgej]

Can you tel me how are you going to use those features

[lvjunmei]

thanks..first ,i have user's  age and sex and rating,i don't know how to add this to the input .because the example of Movielens only use user id,item id and timestamp,so i understand only like rating matrix can be a input.
and i want to know how to use checkpoint(gl.500nc.nc) continue to train save as gl.nc . thank you.

[rgeorgej]

The examples which we gave you had only feature which was the purchase. But you want to add 3 Features for each customer . Am i Right with that ?

[lvjunmei]

yes,right

[spacelover1]

Hi,
Okay so no answers provided here and I'm having the same question, that how can I add more features of either items or users, to improve the recommendations?

[scottlegrand]

If you look at the CIFAR-10 sample, it shows how to build your own training and inference pipeline.

If that's too complicated, Amazon to the best of my knowledge has deprecated this framework and will not be throwing additional resources at supporting it, so I would focus on making use of TF or PyTorch instead, pressing them for better sparse data performance because TF has made zero progress in that direction since 2016 and PyTorch could use some sparse TLC as well.