List your company

Mathematical Reasoning on (GSM8K, MATH, MathQA, MathPile) (test)

49.36GSM8K AccuracyOpT-DeUS
2.518414.679226.8439.0008Dec 15, 2025

Evaluation results

MethodLinks
OpT-DeUSBackbone=Llama-3.1-8B, Shots=5-shot2025.1249.3650.0413.8442.5851.94
MIDUS-HMLBackbone=Llama-3.1-8B, Shots=5-shot2025.1248.951.7814.3442.5852.53
Llama ProBackbone=Llama-3.1-8B, Shots=5-shot2025.1241.9341.7713.140.8745.89
BaseBackbone=Llama-3.1-8B, Shots=5-shot2025.1235.3333.3611.4237.9939.37
MIDUS-HMLBackbone=Llama-3.2-1B, Shots=5-shot2025.126.66.65.6231.3916.73
OpT-DeUSBackbone=Llama-3.2-1B, Shots=5-shot2025.126.076.225.830.9516.35
Llama ProBackbone=Llama-3.2-1B, Shots=5-shot2025.125.614.45.6430.6215.42
BaseBackbone=Llama-3.2-1B, Shots=5-shot2025.124.324.935.6629.8814.93