search

THUDM / GATNE

Source code and dataset for KDD 2019 paper "Representation Learning for Attributed Multiplex Heterogeneous Network"
MIT License
527 stars 141 forks source link

questions for training with GPU #82

Closed sudongxiang closed 3 years ago

sudongxiang commented 3 years ago

Firstly, I have to say, nice work of you, for GATNE is a beautiful model. However, when I start training my own job, training cost too much time. I am wondering, how to solve the problem of low speed in your production environment. Is using GPU the key to solving this problem？

cenyk1230 commented 3 years ago

Hi @sudongxiang,

Thank you for your interest in our work. It really needs GPU for speed-up model training.