List your company

General Reasoning on NQ (Accuracy)

45.67Accuracy (NQ General Reasoning)COSE
14.813222.824130.83538.8459May 27, 2026

Evaluation results

MethodLinks
COSEModel=Qwen2.5-3B-Instruct2026.0545.67
COSEModel=Llama-3.2-3B-Instruct2026.0541.06
COSEModel=Qwen3-4B2026.0540.25
COSEModel=Qwen3-0.6B2026.0539.75
MAEModel=Qwen2.5-3B-Instruct2026.0537.98
AZRModel=Llama-3.2-3B-Instruct2026.0537.26
AZRModel=Qwen3-4B2026.0536.2
R-ZeroModel=Qwen2.5-3B-Instruct2026.0536.1
R-ZeroModel=Qwen3-4B2026.0535.8
MAEModel=Qwen3-4B2026.0535.4
AZRModel=Qwen2.5-3B-Instruct2026.0535.33
MAEModel=Llama-3.2-3B-Instruct2026.0535
R-ZeroModel=Llama-3.2-3B-Instruct2026.0532.78
BaseModel=Qwen3-4B2026.0528
BaseModel=Qwen2.5-3B-Instruct2026.0527.8
BaseModel=Llama-3.2-3B-Instruct2026.0527
MAEModel=Qwen3-0.6B2026.0520.5
R-ZeroModel=Qwen3-0.6B2026.0519.8
AZRModel=Qwen3-0.6B2026.0518.04
BaseModel=Qwen3-0.6B2026.0516