Mathematical Reasoning on GSM8K (accuracy, intervention rate)
87AccuracyAligned base model
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Aligned base modelModel Pair=Q→L, Pair Family=Cross-Family2026.04 | 87 | — | |
| Aligned base modelModel Pair=G→L, Pair Family=Cross-Family2026.04 | 87 | — | |
| Aligned base modelModel Pair=L→L, Pair Family=Within-Family2026.04 | 87 | — | |
| Aligned base modelModel Pair=Q→G, Pair Family=Cross-Family2026.04 | 86 | — | |
| Aligned base modelModel Pair=L→G, Pair Family=Cross-Family2026.04 | 86 | — | |
| Aligned base modelModel Pair=G→G, Pair Family=Within-Family2026.04 | 86 | — | |
| NUDGINGModel Pair=L→G, Pair Family=Cross-Family2026.04 | 67 | 9.2 | |
| BlendInModel Pair=L→G, Pair Family=Cross-Family2026.04 | 67 | 9.2 | |
| NUDGINGModel Pair=G→G, Pair Family=Within-Family2026.04 | 66 | 10 | |
| BlendInModel Pair=G→G, Pair Family=Within-Family2026.04 | 66 | 10 | |
| BlendInModel Pair=G→L, Pair Family=Cross-Family2026.04 | 60 | 14.8 | |
| NUDGINGModel Pair=G→L, Pair Family=Cross-Family2026.04 | 59 | 14.8 | |
| NUDGINGModel Pair=L→L, Pair Family=Within-Family2026.04 | 58 | 14.4 | |
| BlendInModel Pair=L→L, Pair Family=Within-Family2026.04 | 58 | 14.4 | |
| BlendInModel Pair=Q→G, Pair Family=Cross-Family2026.04 | 56 | 23.1 | |
| NUDGINGModel Pair=Q→G, Pair Family=Cross-Family2026.04 | 54 | 23.1 | |
| Guidance ModelModel Pair=Q→L, Pair Family=Cross-Family2026.04 | 53 | — | |
| Guidance ModelModel Pair=Q→G, Pair Family=Cross-Family2026.04 | 53 | — | |
| Guidance ModelModel Pair=G→L, Pair Family=Cross-Family2026.04 | 45 | — | |
| Guidance ModelModel Pair=L→G, Pair Family=Cross-Family2026.04 | 45 | — | |
| Guidance ModelModel Pair=L→L, Pair Family=Within-Family2026.04 | 45 | — | |
| Guidance ModelModel Pair=G→G, Pair Family=Within-Family2026.04 | 45 | — | |
| Base ModelModel Pair=Q→G, Pair Family=Cross-Family2026.04 | 32 | — | |
| Base ModelModel Pair=L→G, Pair Family=Cross-Family2026.04 | 32 | — | |
| Base ModelModel Pair=G→G, Pair Family=Within-Family2026.04 | 32 | — | |
| BlendInModel Pair=Q→L, Pair Family=Cross-Family2026.04 | 31 | 22.2 | |
| NUDGINGModel Pair=Q→L, Pair Family=Cross-Family2026.04 | 27 | 22.2 | |
| Base ModelModel Pair=Q→L, Pair Family=Cross-Family2026.04 | 11 | — | |
| Base ModelModel Pair=G→L, Pair Family=Cross-Family2026.04 | 11 | — | |
| Base ModelModel Pair=L→L, Pair Family=Within-Family2026.04 | 11 | — |