List your company

Mathematical Reasoning on GSM8K (Acc, FLOPs, A/C)

94.6Accuracy (GSM8K)7B Only
84.82487.36289.992.438May 7, 2026

Evaluation results

MethodLinks
7B OnlyExternal Models=N/A2026.0594.64.4221.4
OursExternal Models=false2026.0594.52.0346.55
RSDExternal Models=true2026.0594.42.2741.59
STEERExternal Models=false2026.0594.42.6635.49
1.5B OnlyExternal Models=N/A2026.0585.20.97387.56
SpecReasonExternal Models=false2026.0585.21.0382.72