List your company

General Numerical Reasoning on DROP (Accuracy (%))

82.55AccuracyVeriTime
52.85860.566568.27575.9835Feb 8, 2026

Evaluation results

MethodLinks
VeriTimeBackbone=Qwen2.5-3B-Instruct2026.0282.55
VeriTimeBackbone=Qwen3-4B-Instruct2026.0280
Qwen3-4B-InstructSetting=Base2026.0273.83
Qwen2.5-3B-InstructSetting=Base2026.0254