List your company

Commonsense Reasoning on PIQA, HellaSwag, WinoGrande, ARC

79.4PIQA AccuracyFP16
78.56878.7847979.216Mar 18, 2026

Evaluation results

MethodLinks
FP16Backbone=Llama-2-7B2026.0379.458.26453.963.9
RAMPBackbone=Llama-2-7B2026.0379.257.963.753.663.6
AWQ-4Backbone=Llama-2-7B2026.0378.857.162.352.462.7
Q4_K_MBackbone=Llama-2-7B2026.0378.656.962.152.162.4