List your company

Question Answering with Abstention on ABSTAIN-QA

83.4Accuracy (A)DeepSeek-R1
51.6859.91568.1576.385Apr 18, 2026

Evaluation results

MethodLinks
DeepSeek-R1Model=DeepSeek-R12026.0483.40.29.1
DeepSeek-V3Model=DeepSeek-V32026.0477.54.331.2
Qwen2.5 32B InstructModel=Qwen2.5 32B Instruct2026.0470.78.534.7
Llama3.1 8B InstructModel=Llama3.1 8B Instruct2026.0459.8110.3
Qwen2.5 7B InstructModel=Qwen2.5 7B Instruct2026.0458.98.835.5
Abstain-R1Model=Abstain-R12026.0453.316.840.1
Qwen2.5 3B InstructModel=Qwen2.5 3B Instruct2026.0452.915.330