Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Zero-shot QA and Reasoning on Downstream Suite (PIQA, HS, ARC, WG, RTE, OQA, BoolQ)

80.79PIQA Accuracy

Meta-Llama-3-8B Dense

71.346873.798476.2578.7016Mar 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
80.7979.1777.6953.3372.8569.684581.4469.99
2026.03
79.6576.0775.8447.7871.9870.444.280.768.33
2026.03
79.5476.6974.7548.7272.6967.874380.2467.94
2026.03
79.1175.887549.473.3263.1843.878.3267.25
2026.03
78.7876.2976.5251.1973.0960.6540.281.6267.29
2026.03
78.475.3273.9949.1573.858.8441.879.4266.34
2026.03
74.5963.6660.4836.2660.6956.6837.263.9856.69
2026.03
73.8360.4258.0434.4760.3853.7936.858.254.49
2026.03
73.2360.6659.2634.5661.0159.2136.664.1356.08
2026.03
72.9159.0456.8233.1158.8857.763557.0353.82
2026.03
72.859.9257.6233.1158.6455.9635.259.6654.11
2026.03
71.7157.9657.2833.7959.9852.713363.9453.8