List your company

Question Answering on GPQA (Accuracy, Avg Tokens)

60AccuracyMARS
28.456836.645944.83553.0241Sep 24, 2025

Evaluation results

MethodLinks
MARSBackbone=Llama3-70b2025.09606,633
MADBackbone=Llama3-70b2025.0959.6719,255
Self-consistencyBackbone=Llama3-70b2025.0956.673,470
Self-reflectionBackbone=Llama3-70b2025.0956.332,079
DMADBackbone=Llama3-70b2025.0956.3318,082
MARSBackbone=GPT-4o-mini2025.0948.337,903
CoTBackbone=Llama3-70b2025.0947.67963
MADBackbone=GPT-4o-mini2025.0947.517,083
DMADBackbone=GPT-4o-mini2025.0947.3317,229
MADBackbone=Mixtral 8x22b2025.094713,308
MARSBackbone=Mixtral 8x22b2025.09445,418
CoTBackbone=GPT-4o-mini2025.0943.33677
DMADBackbone=Mixtral 8x22b2025.0943.3313,509
Self-consistencyBackbone=Mixtral 8x22b2025.0941.672,840
CoTBackbone=Mixtral 8x22b2025.0941.33721
Self-consistencyBackbone=GPT-4o-mini2025.0940.672,910
Self-reflectionBackbone=Mixtral 8x22b2025.09381,579
Self-reflectionBackbone=GPT-4o-mini2025.0937.671,597
MARSBackbone=Mixtral 8x7b2025.0937.336,441
MARSBackbone=GPT-3.52025.0936.333,741
Self-consistencyBackbone=Mixtral 8x7b2025.0935.673,033
MADBackbone=Mixtral 8x7b2025.0935.6712,835
DMADBackbone=GPT-3.52025.0935.675,609
Self-reflectionBackbone=Mixtral 8x7b2025.09351,625
DMADBackbone=Mixtral 8x7b2025.093414,139
CoTBackbone=Mixtral 8x7b2025.0933.67759
Self-consistencyBackbone=GPT-3.52025.0933.671,686
CoTBackbone=GPT-3.52025.0931414
MADBackbone=GPT-3.52025.09317,567
Self-reflectionBackbone=GPT-3.52025.0929.67890