List your company
ResearchFormerly WizWand

General Knowledge Reasoning on MMLU (Accuracy)

54.6AccuracySupermajority Con.
42.53645.66848.851.932Jul 6, 2026
Updated 3mo ago

Evaluation results

MethodLinks
Supermajority Con.Backbone=Llama 8B, Protocol=Supermajority Consensus2026.0754.6
Unanimity Con.Backbone=Llama 8B, Protocol=Unanimity Consensus2026.0754.2
JudgeBackbone=Llama 8B, Protocol=Judge2026.0753.7
Simple VotingBackbone=Llama 8B, Protocol=Simple Voting2026.0753.3
Majority ConsensusBackbone=Llama 8B, Protocol=Majority Consensus2026.0753.2
Baseline with CoTBackbone=Llama 8B, Protocol=Baseline with Chain of Thought2026.0753.1
Cumulative VotingBackbone=Llama 8B, Protocol=Cumulative Voting2026.0752.6
Ranked VotingBackbone=Llama 8B, Protocol=Ranked Voting2026.0749.2
BaselineBackbone=Llama 8B, Protocol=Standard Baseline2026.0744.8
Approval VotingBackbone=Llama 8B, Protocol=Approval Voting2026.0743