Skip to content
Branch: master
Find file Copy path
Find file Copy path
1 contributor

Users who have contributed to this file

22 lines (14 sloc) 733 Bytes

PinSage model

NOTE: this version is not using NodeFlow yet.

First, download and extract from

One can then run the following to train PinSage on MovieLens-1M:

python3 --opt Adam --lr 1e-3 --sched none --sgd-switch 25

One can also incorporate user and movie features into training:

python3 --opt Adam --lr 1e-3 --sched none --sgd-switch 25 --use-feature

Currently, performance of PinSage on MovieLens-1M has the best mean reciprocal rank of 0.032298±0.048078 on validation (and 0.033695±0.051963 on test set for the same model). The Implicit Factorization Model from Spotlight has a 0.034572±0.041653 on the test set.

You can’t perform that action at this time.