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

Zero-shot Reasoning on Qwen3 Evaluation Suite (PIQA, ARC-E, ARC-C, HellaSwag, Winogrande, MMLU)

79.76PIQA Accuracy

bf16

74.445675.825377.20578.5847Jun 24, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
79.7682.8760.2478.8272.8577.375.31
2026.06
79.6581.4858.1977.7672.3874.9474.07
2026.06
79.3881.7859.6477.572.0676.0174.4
2026.06
79.3379.925778.0670.4875.3773.36
2026.06
77.6980.9356.5774.9367.5672.9671.77
2026.06
76.8279.3455.0372.7466.6170.3270.14
2026.06
76.3979.3856.472.8766.0669.4370.09
2026.06
74.6575.8452.6572.9364.8868.6468.27