List your company

Language Understanding on MMLU (Overall Accuracy)

79.9MMLU AccuracySFT
50.88458.41765.9573.483May 21, 2025

Evaluation results

MethodLinks
SFTBackbone=Qwen2.5-32B2025.0579.9
RawBackbone=Qwen2.5-32B2025.0579.6
RawBackbone=Qwen2.5-7B2025.0571.3
SFTBackbone=Qwen2.5-7B2025.0571.3
GRPOBackbone=Qwen2.5-7B2025.0571.3
SFTBackbone=Llama3.1-8B2025.0561.4
GRPOBackbone=Llama3.1-8B2025.0561.1
RawBackbone=Llama3.1-8B2025.0560.4
RawBackbone=Llama3.2-3B2025.0552.7
SFTBackbone=Llama3.2-3B2025.0552.1
GRPOBackbone=Llama3.2-3B2025.0552