Non-Arithmetic Reasoning on FAMMA Non-Arithmetic (test)
58.74AccuracyASDA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ASDAStudent=Haiku 4.5, Teacher Model=Sonnet 4.5, Stage=Best refinement epoch (E2)2026.03 | 58.74 | 1.6 | |
| BaselineStudent=Haiku 4.5, Teacher Model=Sonnet 4.52026.03 | 57.14 | — | |
| ASDAStudent=Haiku 4.5, Teacher Model=Sonnet 4.5, Stage=Warm-Up (WU)2026.03 | 56.35 | 0.79 | |
| ASDAStudent=Haiku 3.5, Teacher Model=Sonnet 4.5, Stage=Best refinement epoch (E2)2026.03 | 55.16 | 5.95 | |
| ASDAStudent=Haiku 3.5, Teacher Model=Sonnet 4.5, Stage=Warm-Up (WU)2026.03 | 51.98 | 2.78 | |
| GEPAStudent=Haiku 3.5, Teacher Model=Sonnet 4.52026.03 | 50.79 | 1.58 | |
| ACEStudent=Haiku 3.5, Teacher Model=Sonnet 4.52026.03 | 49.6 | 0.39 | |
| BaselineStudent=Haiku 3.5, Teacher Model=Sonnet 4.52026.03 | 49.21 | — |