List your company
ResearchFormerly WizWand

Commonsense Reasoning on MS

93.2First-Token AccuracyLlama-3.1-8B
63.14470.94778.7586.553May 21, 2025
Updated 7mo ago

Evaluation results

MethodLinks
Llama-3.1-8BStrategy=Prefilling2025.0593.2
Phi-4-14BStrategy=Prefilling2025.0593.2
Phi-4-14BStrategy=Prompt Engineering2025.0593.1
Qwen-2-7BStrategy=Prefilling2025.0592.6
Gemma-2-9BStrategy=Prompt Engineering2025.0591.6
Llama-3.1-8BStrategy=Prompt Engineering2025.0590.1
Ministral-8BStrategy=Prefilling2025.0589.9
Qwen-2-7BStrategy=FTP2025.0588.1
Gemma-2-9BStrategy=Prefilling2025.0587.3
Ministral-8BStrategy=FTP2025.0587.3
Llama-3.1-8BStrategy=FTP2025.0587
Zephyr-7BStrategy=Prefilling2025.0586.2
Gemma-2-9BStrategy=FTP2025.0585.9
Ministral-8BStrategy=Prompt Engineering2025.0585.7
Qwen-2-7BStrategy=Prompt Engineering2025.0585.3
Zephyr-7BStrategy=Prompt Engineering2025.0585.3
Zephyr-7BStrategy=FTP2025.0583.6
Phi-4-14BStrategy=FTP2025.0583.4
Mistral-Nemo-12BStrategy=Prefilling2025.0580.9
Mistral-Nemo-12BStrategy=FTP2025.0578
Gemma-7BStrategy=FTP2025.0577.5
Mistral-Nemo-12BStrategy=Prompt Engineering2025.0574.2
Gemma-7BStrategy=Prompt Engineering2025.0565.7
Gemma-7BStrategy=Prefilling2025.0564.3