Mathematical Reasoning on GSM8K matched-N groups (test)
96.33AccuracyOracle topology
Evaluation Results
| Method | Links | |
|---|---|---|
| Oracle topologyTopology Category=Confidence-routed2026.05 | 96.33 | |
| hub-spokeTopology Category=Fixed topology2026.05 | 95.33 | |
| Per-bench best fixedTopology Category=Fixed topology2026.05 | 95.33 | |
| Majority over topologiesTopology Category=Selection w/o learned confidence2026.05 | 95.33 | |
| Highest plurality shareTopology Category=Selection w/o learned confidence2026.05 | 95.33 | |
| CAGE-SELECTTopology Category=Confidence-routed2026.05 | 95.33 | |
| treeTopology Category=Fixed topology2026.05 | 95 | |
| iidTopology Category=Fixed topology2026.05 | 94.67 | |
| debateTopology Category=Fixed topology2026.05 | 94.67 | |
| Highest mean log-probTopology Category=Selection w/o learned confidence2026.05 | 94.67 | |
| chainTopology Category=Fixed topology2026.05 | 94 |