Tool-Integrated Reasoning on AIME 25 (test accuracy)
31.25Test AccuracyToken-ALP
Evaluation Results
| Method | Links | |
|---|---|---|
| Token-ALPtemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 31.25 | |
| Token-MIStemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 28.85 | |
| Seq-ALPtemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 28.65 | |
| GRPOtemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 24.79 | |
| Seq-MIStemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 24.17 | |
| Seq-Bypasstemperature=1.0, sampling=average@32, multi-turn=true2026.03 | 24.06 |