Mathematical Reasoning on AIME 2024-2026
2.47AccuracyGRPO + RSI-S
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GRPO + RSI-SBackbone=Qwen2.5-1.5B, Evaluation Protocol=avg@322026.06 | 2.47 | 1,082.7 | 0.11 | -137.95 | |
| GRPOBackbone=Qwen2.5-1.5B, Evaluation Protocol=avg@322026.06 | 2.36 | 1,220.65 | — | — | |
| GRPO + EBBackbone=Qwen2.5-1.5B, Evaluation Protocol=avg@322026.06 | 2.22 | 1,138.05 | — | — | |
| GRPO + PBBackbone=Qwen2.5-1.5B, Evaluation Protocol=avg@322026.06 | 2.12 | 1,084.09 | — | — |