ResearchBenchmarksMulti-task Language Understanding on MMLU (0-shot)Follow69.11Exact Match (EM)VAR30.983640.881850.7860.6782Feb 16, 2025Evaluation ResultsMethodMethodLinksExact Match (EM)VARBase Model=Qwen2.5-7BBase Model=Qwen2.5-7B2025.0269.11DPOBase Model=Qwen2.5-7BBase Model=Qwen2.5-7B2025.0268.64ALoLBase Model=Qwen2.5-7BBase Model=Qwen2.5-7B2025.0268.62BaseBase Model=Qwen2.5-7BBase Model=Qwen2.5-7B2025.0267.13VARBase Model=Llama2-7BBase Model=Llama2-7B2025.0238.57BaseBase Model=Llama2-7BBase Model=Llama2-7B2025.0237.46ALoLBase Model=Llama2-7BBase Model=Llama2-7B2025.0235.78DPOBase Model=Llama2-7BBase Model=Llama2-7B2025.0232.45