Table 4 Accuracy(%) and parameter counts.
From: GTAT: empowering graph neural networks with cross attention
Model | Cora | Photo | Physics | |||
|---|---|---|---|---|---|---|
#Param | ACC | #Param | ACC | #Param | ACC | |
GAT | 184K | 79.88±1.38 | 96K | 90.33±1.58 | 1078K | 92.90±1.12 |
GTAT | 187K | 80.34±0.97 | 99K | 91.07±0.65 | 1080K | 93.13±0.65 |
GAT2 | 184K | 80.21±1.35 | 96K | 90.89±1.20 | 1078K | 92.29±0.79 |
GTAT2 | 187K | 80.84±1.26 | 99K | 91.17±1.01 | 1080K | 92.93±0.82 |
