List your company

Mathematical Reasoning on GSM8K (accuracy, intervention rate)

87AccuracyAligned base model
7.9628.484969.52Apr 22, 2026

Evaluation results

MethodLinks
Aligned base modelModel Pair=Q→L, Pair Family=Cross-Family2026.0487—
Aligned base modelModel Pair=G→L, Pair Family=Cross-Family2026.0487—
Aligned base modelModel Pair=L→L, Pair Family=Within-Family2026.0487—
Aligned base modelModel Pair=Q→G, Pair Family=Cross-Family2026.0486—
Aligned base modelModel Pair=L→G, Pair Family=Cross-Family2026.0486—
Aligned base modelModel Pair=G→G, Pair Family=Within-Family2026.0486—
NUDGINGModel Pair=L→G, Pair Family=Cross-Family2026.04679.2
BlendInModel Pair=L→G, Pair Family=Cross-Family2026.04679.2
NUDGINGModel Pair=G→G, Pair Family=Within-Family2026.046610
BlendInModel Pair=G→G, Pair Family=Within-Family2026.046610
BlendInModel Pair=G→L, Pair Family=Cross-Family2026.046014.8
NUDGINGModel Pair=G→L, Pair Family=Cross-Family2026.045914.8
NUDGINGModel Pair=L→L, Pair Family=Within-Family2026.045814.4
BlendInModel Pair=L→L, Pair Family=Within-Family2026.045814.4
BlendInModel Pair=Q→G, Pair Family=Cross-Family2026.045623.1
NUDGINGModel Pair=Q→G, Pair Family=Cross-Family2026.045423.1
Guidance ModelModel Pair=Q→L, Pair Family=Cross-Family2026.0453—
Guidance ModelModel Pair=Q→G, Pair Family=Cross-Family2026.0453—
Guidance ModelModel Pair=G→L, Pair Family=Cross-Family2026.0445—
Guidance ModelModel Pair=L→G, Pair Family=Cross-Family2026.0445—
Guidance ModelModel Pair=L→L, Pair Family=Within-Family2026.0445—
Guidance ModelModel Pair=G→G, Pair Family=Within-Family2026.0445—
Base ModelModel Pair=Q→G, Pair Family=Cross-Family2026.0432—
Base ModelModel Pair=L→G, Pair Family=Cross-Family2026.0432—
Base ModelModel Pair=G→G, Pair Family=Within-Family2026.0432—
BlendInModel Pair=Q→L, Pair Family=Cross-Family2026.043122.2
NUDGINGModel Pair=Q→L, Pair Family=Cross-Family2026.042722.2
Base ModelModel Pair=Q→L, Pair Family=Cross-Family2026.0411—
Base ModelModel Pair=G→L, Pair Family=Cross-Family2026.0411—
Base ModelModel Pair=L→L, Pair Family=Within-Family2026.0411—