The authors start from existing graph metanetworks (GMNs) and apply three modifications: fan-in rescaling of input weights, mean aggregation in message passing, and a layer-wise mean readout.
Graph metanetworks can generalize across widths with principled modifications
The authors propose Transferable Graph Metanetworks, achieving size transfer by enforcing invariance and continuity.
Big Tech
Yuxin Ma · Adir Dayan · Yam Eitan · Haggai Maron · Soledad Villar
Johns Hopkins University · Technion · NVIDIA
Research Digest··2 min read
The authors introduce modifications to graph metanetworks that make predictions transferable across input networks of different widths.
Why this paper
From NVIDIA and 2 others
In one line
Graph metanetworks can be modified to generalize across input networks of different widths, enabling training on small networks and evaluation on larger ones.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (3 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§