List your company

Mathematical Reasoning on GSM8k-Aug (rationale format original)

98.89Pass@1Ours-RL
20.068440.531760.99581.4583May 25, 2026

Evaluation results

MethodLinks
Ours-RLRationale Format=GSM8k-Aug2026.0598.895.68
SFT-CoTRationale Format=GSM8k-Aug2026.0597.2214.3
Ours-SFTRationale Format=GSM8k-Aug2026.05955.73
CoLaR-2Rationale Format=GSM8k-Aug2026.0591.37.35
SFT-CoTRationale Format=GSM8k-Aug2026.0552.7725.2
Ours-RLRationale Format=GSM8k-Aug2026.0547.2313.23
Ours-SFTRationale Format=GSM8k-Aug2026.0542.4613.42
CoconutRationale Format=GSM8k-Aug2026.0541.16
CoLaR-2Rationale Format=GSM8k-Aug2026.0540.112.7
CoconutRationale Format=GSM8k-Aug2026.0523.16