Polynomial-Objective Integer Programming on RandQCP 2k
0.31Gap (%)HNN
Evaluation results
| Method | Links | |
|---|---|---|
| HNNTrain=Mini, Base solver=Gurobi2026.03 | 0.31 | |
| HNNTrain=1000, Base solver=Gurobi2026.03 | 0.32 | |
| HNNTrain=Mini, Base solver=SCIP2026.03 | 0.37 | |
| HNNTrain=2000, Base solver=Gurobi2026.03 | 0.38 | |
| NeuralQPTrain=2000, Base solver=Gurobi2026.03 | 0.39 | |
| HNNTrain=2000, Base solver=SCIP2026.03 | 0.4 | |
| NeuralQPTrain=1000, Base solver=Gurobi2026.03 | 0.42 | |
| HNNTrain=1000, Base solver=SCIP2026.03 | 0.42 | |
| NeuralQPTrain=Mini, Base solver=Gurobi2026.03 | 0.43 | |
| TriGNNTrain=Mini, Base solver=Gurobi2026.03 | 0.45 | |
| NeuralQPTrain=2000, Base solver=SCIP2026.03 | 0.47 | |
| TriGNNTrain=1000, Base solver=Gurobi2026.03 | 0.47 | |
| TriGNNTrain=2000, Base solver=SCIP2026.03 | 0.47 | |
| TriGNNTrain=Mini, Base solver=SCIP2026.03 | 0.49 | |
| NeuralQPTrain=1000, Base solver=SCIP2026.03 | 0.5 | |
| TriGNNTrain=2000, Base solver=Gurobi2026.03 | 0.5 | |
| NeuralQPTrain=Mini, Base solver=SCIP2026.03 | 0.51 | |
| TriGNNTrain=1000, Base solver=SCIP2026.03 | 0.53 | |
| GNNQPTrain=Mini, Base solver=Gurobi2026.03 | 2.29 | |
| GNNQPTrain=1000, Base solver=Gurobi2026.03 | 2.3 | |
| GNNQPTrain=2000, Base solver=Gurobi2026.03 | 2.32 | |
| GNNQPTrain=2000, Base solver=SCIP2026.03 | 2.37 | |
| GNNQPTrain=Mini, Base solver=SCIP2026.03 | 2.4 | |
| GNNQPTrain=1000, Base solver=SCIP2026.03 | 2.4 | |
| Exact solverTrain=-, Base solver=Gurobi2026.03 | 6.45 | |
| Exact solverTrain=-, Base solver=SCIP2026.03 | 42.08 |