Papers12:03Chinese content重新审视关系深度学习简单基线：特征传播加MLP逼近异质GNN做关系库预测先跑这条基线，MLP加特征传播在RelBench上就能逼近GNN，省算力。#RelBench#图神经网络#关系深度学习#Correct-and-SmoothaarXiv cs.LG@Viet The Nguyen et al. (3)SourceRead laterReadWorth followingUsefulFollow RelBench