List your company
ResearchFormerly WizWand

Multi-task Language Understanding on MMLU (Inference Latency)

2.07Average Inference Time (s)CoT
-2.08225.94453.9781.996Sep 24, 2025
Updated 7mo ago

Evaluation results

MethodLinks
CoTModel=GPT-3.52025.092.07
CoTModel=Mixtral 8x22b2025.092.29
CoTModel=Llama3-70b2025.092.73
CoTModel=GPT-4o-mini2025.093.48
Self-reflectionModel=Mixtral 8x22b2025.093.9
Self-reflectionModel=GPT-3.52025.094.28
Self-consistencyModel=GPT-3.52025.094.93
Self-reflectionModel=GPT-4o-mini2025.095.42
Self-reflectionModel=Llama3-70b2025.096.86
MARSModel=GPT-3.52025.097.61
DMADModel=GPT-3.52025.0910.67
MADModel=GPT-3.52025.0911.38
Self-consistencyModel=Mixtral 8x22b2025.0914.44
Self-consistencyModel=GPT-4o-mini2025.0914.55
MARSModel=Mixtral 8x22b2025.0916.15
MADModel=Mixtral 8x22b2025.0917.39
CoTModel=Mixtral 8x7b2025.0920.38
DMADModel=Mixtral 8x22b2025.0923.89
MARSModel=GPT-4o-mini2025.0928.14
Self-consistencyModel=Llama3-70b2025.0929.6
Self-reflectionModel=Mixtral 8x7b2025.0929.82
MARSModel=Llama3-70b2025.0936.49
MADModel=GPT-4o-mini2025.0941.58
DMADModel=GPT-4o-mini2025.0942.96
Self-consistencyModel=Mixtral 8x7b2025.0947.76
MADModel=Llama3-70b2025.0951.78
MARSModel=Mixtral 8x7b2025.0955.91
DMADModel=Llama3-70b2025.0964.03
MADModel=Mixtral 8x7b2025.0977.31
DMADModel=Mixtral 8x7b2025.09105.87