List your company

Reasoning and Knowledge Assessment on Xstorycloze bo

72.96AccuracyOurs-MoE-SFT
47.521654.125860.7367.3342Jul 12, 2025

Evaluation results

MethodLinks
Ours-MoE-SFTModel=Ours-MoE-SFT2025.0772.96
Ours-SFTModel=Ours-SFT2025.0761.86
Ours-MoE-Base-8kModel=Ours-MoE-Base-8k2025.0760.86
Ours-BaseModel=Ours-Base2025.0760.8
Ours-MoE-BaseModel=Ours-MoE-Base2025.0760.74
Ours-Base-32kModel=Ours-Base-32k2025.0760.6
LLaMA3.1-8B-InstructModel=LLaMA3.1-8B-Instruct2025.0751.69
Qwen2.5-7B-InstructModel=Qwen2.5-7B-Instruct2025.0750.43
Qwen3-8BModel=Qwen3-8B2025.0750.37
Qwen2.5-7B-baseModel=Qwen2.5-7B-base2025.0749.97
DeepSeek-R1-Distill-Llama-8BModel=DeepSeek-R1-Distill-Llama-8B2025.0748.5