Next-token Reasoning on OMNI-MATH Medium (val)
61.15Accuracy (Next-token Reasoning)LoopRPT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LoopRPTModel=Ouro-2.6B, Reasoning Mode=Peak2026.03 | 61.15 | 4 | |
| LoopRPTModel=Ouro-2.6B, Reasoning Mode=Adap.2026.03 | 60.21 | 2.18 | |
| LoopRPTModel=Ouro-1.4B, Reasoning Mode=Peak2026.03 | 58.29 | 4 | |
| LoopRPTModel=Ouro-1.4B, Reasoning Mode=Adap.2026.03 | 57.72 | 2.81 | |
| Adap.Model=Ouro-2.6B2026.03 | 57.35 | 3.35 | |
| PeakModel=Ouro-1.4B2026.03 | 57.28 | 4 | |
| Adap.Model=Ouro-1.4B2026.03 | 57.2 | 3.53 | |
| PeakModel=Ouro-2.6B2026.03 | 57.19 | 4 | |
| VanillaModel=Qwen3-1.7B2026.03 | 32.18 | — | |
| +CoTModel=Qwen3-1.7B2026.03 | 8.6 | — |