An implementation of the paper Order Matters: Sequence to Sequence for Sets.
git clone https://github.com/arunppsg/Set2Set.git & cd set2set
python setup.py install
import torch
import set2set as s2s
X = torch.randn(2, 5, 10)
model = s2s.Set2Set(10, 3, 1)
embedding = model(X)