List your company

Mathematical Reasoning on GSM8K (EM Strict/Flex)

38.89EM (Strict)Baseline
36.841237.373137.90538.4369May 26, 2026

Evaluation results

MethodLinks
BaselineEvaluation Framework=lm-eval-harness2026.0538.8939.04
ReMoEEvaluation Framework=lm-eval-harness2026.0538.1338.36
CE-onlyEvaluation Framework=lm-eval-harness, Training=Generic router fine-tuning2026.0536.9237.23