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

Commonsense Reasoning on ARC-e, OBQA, SIQA, ARC-c, WinoG, PIQA, BoolQ, HellaS (LLaMA3-8B)

92.9ARC-e Accuracy

EffiLoRA

83.85286.20188.5590.899Nov 30, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.11
92.98781.781.888.489.674.195.886.4
2025.11
92.48782.681.987.88873.696.286.1
2025.11
90.48576.779.183.386.475.175.481.4
2025.11
90.388.279.480.683.788.875.995.385.3
2025.11
89.886.680.579.984.488.372.794.784.6
2025.11
89.281.878.776.782.181.665.19280.9
2025.11
88.28579.177.584.386.473.394.183.1
2025.11
87.88574.879.583.487.172.494.883.5
2025.11
87.583.180.2758485.57196.182.8
2025.11
86.584.878.879.982.187.67593.383.5
2025.11
86.484.476.477.983.386.77493.982.9
2025.11
86.18179.573.483.884.26993.481.3
2025.11
84.27979.971.284.385.270.891.780.8