Algorithmic Reasoning on CLRS-30 n=64 (test)
100Sort AccuracyFloydNet
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FloydNetTraining size=n=16, Testing size=n=642026.01 | 100 | 91.6 | 86.2 | 93.17 | 90.02 | 98.6 | 99.72 | 99.68 | 96.64 | |
| RANRClass=GNN, Training size=n=16, Testing size=n=642026.01 | 94.15 | 82.9 | 83.53 | 91.66 | 42.69 | 74.15 | 49.12 | 88.44 | 75.78 | |
| ETClass=GT, Training size=n=16, Testing size=n=642026.01 | 82.26 | 63 | 64.44 | 81.67 | 83.49 | 86.08 | 54.84 | 88.22 | 80.13 | |
| G-ForgetNetClass=GNN, Training size=n=16, Testing size=n=642026.01 | 78.09 | 63.84 | 78.97 | 91.79 | 86.7 | 88.8 | 54.73 | 95.09 | 82.89 | |
| Triplet-GMPNNClass=GNN, Training size=n=16, Testing size=n=642026.01 | 75.64 | 58.83 | 76.36 | 91.22 | 81.99 | 86.42 | 49.09 | 94.09 | 80.04 | |
| RTClass=GT, Training size=n=16, Testing size=n=642026.01 | 50.01 | 65.31 | 66.52 | 85.33 | 83.2 | 65.33 | 32.52 | 84.55 | 66.18 |