List your company

Question Answering on ARC-C (avg.@8)

92.42Accuracy (ARC-C avg.@8)Qwen2.5-7B-Instruct
87.542488.808790.07591.3413Aug 25, 2025

Evaluation results

MethodLinks
Qwen2.5-7B-InstructTraining Pipeline=PSFT → GRPO2025.0892.42
Qwen2.5-7B-InstructTraining Pipeline=PSFT2025.0892.29
Qwen2.5-7B-InstructTraining Pipeline=SFT → GRPO2025.0892.25
Qwen2.5-7B-InstructTraining Pipeline=SFT2025.0892.22
Llama3.1-8B-InstructTraining Pipeline=PSFT → GRPO2025.0890.92
Llama3.1-8B-InstructTraining Pipeline=PSFT2025.0889.11
Llama3.1-8B-InstructTraining Pipeline=SFT → GRPO2025.0888.57
Llama3.1-8B-InstructTraining Pipeline=SFT2025.0887.73