List your company

Question Answering with Abstention on ABSTAIN (test)

78.6AccuracyDeepSeek-R1
47.60855.65463.771.746Apr 18, 2026

Evaluation results

MethodLinks
DeepSeek-R1Model=DeepSeek-R12026.0478.68.585.952.246.589.1
DeepSeek-V3Model=DeepSeek-V32026.0477.810.987.358.450.185.8
Qwen2.5 32B InstructModel=Qwen2.5 32B Instruct2026.0471.911.581.256.930.754
Qwen2.5 7B InstructModel=Qwen2.5 7B Instruct2026.0462.412.771.547.22.34.9
Llama3.1 8B InstructModel=Llama3.1 8B Instruct2026.0458.38.563.733.800
Abstain-R1Model=Abstain-R12026.0457.220.471.968.155.180.9
Qwen2.5 3B InstructModel=Qwen2.5 3B Instruct2026.0448.818.860.19.40.66.4