Reading Comprehension on SQuAD (F1 Score)
58F1 ScoreRanked Voting
Evaluation Results
| Method | Links | |
|---|---|---|
| Ranked VotingBackbone=Llama 8B2026.07 | 58 | |
| JudgeBackbone=Llama 8B2026.07 | 57.2 | |
| Simple VotingBackbone=Llama 8B2026.07 | 56.2 | |
| Cumulative VotingBackbone=Llama 8B2026.07 | 55.8 | |
| Baseline with CoTBackbone=Llama 8B2026.07 | 53.8 | |
| BaselineBackbone=Llama 8B2026.07 | 52.3 | |
| Approval VotingBackbone=Llama 8B2026.07 | 46.5 | |
| Supermajority Con.Backbone=Llama 8B2026.07 | 44.4 | |
| Unanimity Con.Backbone=Llama 8B2026.07 | 43.4 | |
| Majority ConsensusBackbone=Llama 8B2026.07 | 43.1 |