List your company

Zero-shot Reasoning Suite (PiQA, ARC-E/C, HellaSwag, WinoGrande, BoolQ)

80.7PiQA Accuracy (Zero-shot)FP
47.305655.975364.64573.3147May 18, 2026

Evaluation results

MethodLinks
FPModel=Llama3-8b, W/A=FP, Evaluation Protocol=Zero-shot2026.0580.777.753.779.173.281.174.3
FPModel=Llama2-13b, W/A=FP, Evaluation Protocol=Zero-shot2026.0580.577.549.279.472.480.673.3
ResComp-MARRModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0579.3875.1747.7877.1170.9678.3571.46
FPModel=Llama2-7b, W/A=FP, Evaluation Protocol=Zero-shot2026.057974.646.57668.977.770.5
GPTAQ-MARRModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.7373.9946.8477.2770.4877.4670.79
ResCompModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.6774.4547.8777.170.3280.0371.41
GPTQModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.2474.8346.6776.969.5379.1470.88
GPTAQModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.2473.9146.8477.169.4678.2370.63
GPTAQModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.2471.846.7674.9467.8877.8669.58
ResComp-MARRModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.1874.6246.1674.9767.476.769.67
ResCompModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0578.0272.2247.4474.7267.6476.7969.47
GPTAQ-MARRModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0577.9773.9547.3575.1768.6776.9770.01
ResCompModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0577.8670.4642.6973.7265.2373.0667.17
GPTQModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0577.6971.7642.4973.1765.0475.3267.58
RTNModel=Llama2-13b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0577.3774.0344.873.7368.8276.0969.14
GPTAQ-MARRModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0577.0471.4643.7773.5767.9673.8567.94
ResComp-MARRModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0576.8869.9142.2473.766.0674.7167.25
GPTAQModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0576.570.3741.9873.5265.973.6466.98
GPTQModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0575.7372.1844.2873.8166.6976.1568.14
RTNModel=Llama2-7b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0574.1662.8436.3566.2461.1769.7961.76
RTNModel=Llama3-8b, W/A=W4A4, Evaluation Protocol=Zero-shot2026.0574.165.2840.0270.2665.1974.464.88
GPTAQModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0567.0354.8331.0848.5456.3162.0153.3
GPTAQ-MARRModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0566.3851.3930.3851.9359.0464.1653.88
ResComp-MARRModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0564.7448.2728.7548.6757.6262.9151.83
ResCompModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0563.7651.7626.5547.6554.2260.2950.7
GPTAQModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0562.1344.1525.6842.752.6458.5947.65
GPTAQ-MARRModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0561.4843.925.2643.8457.3861.3148.86
ResComp-MARRModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0560.9443.5625.1742.8155.4962.0548.34
GPTQModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0560.9441.2924.4941.3553.3562.0547.24
ResCompModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0560.7944.1725.0241.7953.4162.0647.87
GPTAQ-MARRModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0559.7939.3525.2640.5854.1458.946.34
GPTAQModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0558.740.4923.2138.9654.2262.2946.31
ResCompModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0558.2538.3223.7438.5453.261.5945.44
ResComp-MARRModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0557.5637.3723.9838.3555.2561.545.67
GPTQModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0556.833.3821.7631.7752.3349.7940.97
GPTQModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0553.4832.9521.2530.0950.7546.1239.11
RTNModel=Llama3-8b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0551.0925.6326.6226.8750.243.1537.26
RTNModel=Llama2-7b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0550.7625.5928.5826.3848.5437.8336.28
RTNModel=Llama2-13b, W/A=W2A4, Evaluation Protocol=Zero-shot2026.0548.5927.1927.1325.0251.7837.8336.26