List your company

Mathematical Reasoning on GSM8K (Accuracy, Rank)

75.2AccuracyBaseline
7.625.1542.760.25May 7, 2026

Evaluation results

MethodLinks
BaselineBackbone=LLaMA-3.1-8B-Instruct2026.0575.2—
SKOPBackbone=LLaMA-3.1-8B-Instruct2026.0566.62.69
Angular SteerBackbone=LLaMA-3.1-8B-Instruct2026.0560.54.75
SADIBackbone=LLaMA-3.1-8B-Instruct2026.0558.63.38
CASTBackbone=LLaMA-3.1-8B-Instruct2026.0541.85.63
LoRABackbone=LLaMA-3.1-8B-Instruct2026.0529—
DISCO-QBackbone=LLaMA-3.1-8B-Instruct2026.0522.54.75
ITIBackbone=LLaMA-3.1-8B-Instruct2026.0520.64.75
CAABackbone=LLaMA-3.1-8B-Instruct2026.05145.44
Comm SteerBackbone=LLaMA-3.1-8B-Instruct2026.0510.24.63