Mathematical Reasoning on AIME25 (ACC, LEN, delta_LEN)
75Accuracy (ACC)Qwen3-32B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3-32BSetting=Steer2026.06 | 75 | — | — | — | |
| Qwen3-14BSetting=Steer2026.06 | 73.33 | — | — | — | |
| Qwen3-32BSetting=Base2026.06 | 69.17 | — | — | — | |
| Qwen3-14BSetting=Base2026.06 | 67.66 | — | — | — | |
| ALRR+ERSSModel=Qwen3-4B2026.06 | 63.33 | — | — | — | |
| SuCoModel Scale=7B2026.06 | 61.7 | — | — | — | |
| Qwen3-8BSetting=Steer2026.06 | 61.67 | — | — | — | |
| ERSSModel=Qwen3-4B2026.06 | 60.83 | — | — | — | |
| Qwen3-4BSetting=Steer2026.06 | 60.82 | — | — | — | |
| Qwen3-8BSetting=Base2026.06 | 60 | — | — | — | |
| ALRRModel=Phi-4-Reasoning2026.06 | 60 | — | — | — | |
| ALRRModel=Qwen3-4B2026.06 | 60 | — | — | — | |
| LRSModel=Qwen3-4B2026.06 | 60 | — | — | — | |
| LLaMA-70BSetting=Steer2026.06 | 58.33 | — | — | — | |
| ALRR+LRSModel=Qwen3-4B2026.06 | 58.33 | — | — | — | |
| S-GRPOModel Scale=7B2026.06 | 58.3 | — | — | — | |
| LHRMsModel Scale=7B2026.06 | 57.7 | — | — | — | |
| LRSModel=Phi-4-Reasoning2026.06 | 56.67 | — | — | — | |
| BaselineModel=Qwen3-4B2026.06 | 56.67 | — | — | — | |
| Qwen3-4BSetting=Base2026.06 | 56.66 | — | — | — | |
| LLaMA-70BSetting=Base2026.06 | 55.83 | — | — | — | |
| AdaCoTModel Scale=7B2026.06 | 55 | — | — | — | |
| AdaptThinkModel Scale=7B2026.06 | 54.3 | — | — | — | |
| CCoTModel=Qwen3-8B2025.10 | 53.33 | 10,438.07 | 17.86 | — | |
| CERModel=Qwen3-8B2025.10 | 53.33 | 9,894.51 | 22.14 | — | |
| ERSSModel=Phi-4-Reasoning2026.06 | 53.33 | — | — | — | |
| CERModel=Qwen3-4B2025.10 | 51.11 | 9,565.64 | 21.07 | — | |
| BaselineModel=Phi-4-Reasoning2026.06 | 50 | — | — | — | |
| ERSS+LRSModel=Qwen3-4B2026.06 | 50 | — | — | — | |
| DeepSeek-R1-DistillModel Scale=7B2026.06 | 49.7 | — | — | — | |
| w thinkingModel=Qwen3-4B2025.10 | 48.89 | 12,119.62 | — | — | |
| w thinkingModel=Qwen3-8B2025.10 | 46.67 | 12,708.16 | — | — | |
| ALRR+ERSSModel=R1-Distilled-LLaMA-8B2026.06 | 43.33 | — | — | — | |
| CCoTModel=Qwen3-4B2025.10 | 40 | 10,775.93 | 11.09 | — | |
| LLaMA-8BSetting=Steer2026.06 | 40 | — | — | — | |
| ERSSModel=R1-Distilled-LLaMA-8B2026.06 | 40 | — | — | — | |
| ALRRModel=R1-Distilled-LLaMA-8B2026.06 | 36.67 | — | — | — | |
| LRSModel=R1-Distilled-LLaMA-8B2026.06 | 36.67 | — | — | — | |
| LHRMsModel Scale=1.5B2026.06 | 34 | — | — | — | |
| SuCoModel Scale=1.5B2026.06 | 33.7 | — | — | — | |
| Qwen3-1.7BSetting=Steer2026.06 | 33.33 | — | — | — | |
| LLaMA-8BSetting=Base2026.06 | 33.33 | — | — | — | |
| BaselineModel=R1-Distilled-LLaMA-8B2026.06 | 33.33 | — | — | — | |
| ALRR+LRSModel=R1-Distilled-LLaMA-8B2026.06 | 33.33 | — | — | — | |
| S-GRPOModel Scale=1.5B2026.06 | 31 | — | — | — | |
| Qwen3-1.7BSetting=Base2026.06 | 30.83 | — | — | — | |
| ALRRModel=Gemma3-12B-it2026.06 | 30 | — | — | — | |
| STRIDEBase Model=Qwen3-4B-Base2026.06 | 28.76 | — | — | 49.24 | |
| ERSSModel=Gemma3-12B-it2026.06 | 28.33 | — | — | — | |
| LRSModel=Gemma3-12B-it2026.06 | 28.33 | — | — | — | |
| AdaptThinkModel Scale=1.5B2026.06 | 28.3 | — | — | — | |
| AdaCoTModel Scale=1.5B2026.06 | 27.3 | — | — | — | |
| CoDModel=Qwen3-4B2025.10 | 26.7 | 3,535.23 | 70.83 | — | |
| DeepSeek-R1-DistillModel Scale=1.5B2026.06 | 26.7 | — | — | — | |
| BaselineModel=Gemma3-12B-it2026.06 | 26.67 | — | — | — | |
| LRSModel=OPEN-REASONER-7B, Prompting=0-shot2026.05 | 26.6 | — | — | — | |
| HICRA-GRPOBase Model=Qwen3-4B-Base2026.06 | 26.32 | — | — | 47.09 | |
| HET-DAPOBase Model=Qwen3-4B-Base2026.06 | 26.12 | — | — | 47.27 | |
| GRPOBase Model=Qwen3-4B-Base2026.06 | 25.67 | — | — | 45.28 | |
| DAPOBase Model=Qwen3-4B-Base2026.06 | 25.34 | — | — | 46.63 | |
| w/o thinkingModel=Qwen3-8B2025.10 | 23.33 | 2,300.13 | 80.69 | — | |
| ERSS+LRSModel=R1-Distilled-LLaMA-8B2026.06 | 23.33 | — | — | — | |
| w/o thinkingModel=Qwen3-4B2025.10 | 20 | 2,503.93 | 82.84 | — | |
| CoDModel=Qwen3-8B2025.10 | 20 | 3,709.2 | 70.81 | — | |
| LRS BASICModel=OPEN-REASONER-7B, Prompting=0-shot2026.05 | 20 | — | — | — | |
| ERSSModel=Phi-42026.06 | 20 | — | — | — | |
| ERSSModel=Gemma3-4B-it2026.06 | 20 | — | — | — | |
| ALRRModel=Phi-42026.06 | 16.67 | — | — | — | |
| ALRRModel=Gemma3-4B-it2026.06 | 16.67 | — | — | — | |
| BaseModel=OPEN-REASONER-7B, Prompting=0-shot2026.05 | 16.6 | — | — | — | |
| LRSModel=Phi-42026.06 | 15.83 | — | — | — | |
| STRIDEBase Model=Qwen2.5-7B-Instruct2026.06 | 15.59 | — | — | 44.71 | |
| LRSModel=Gemma3-4B-it2026.06 | 15 | — | — | — | |
| HET-DAPOBase Model=Qwen2.5-7B-Instruct2026.06 | 13.77 | — | — | 42.18 | |
| BaselineModel=Phi-42026.06 | 13.33 | — | — | — | |
| BaselineModel=Gemma3-4B-it2026.06 | 13.33 | — | — | — | |
| ReSumBackbone=Qwen2.5-Math-7B2026.06 | 13.33 | — | — | — | |
| CoTModel=OPEN-REASONER-7B2026.05 | 13.3 | — | — | — | |
| HICRA-GRPOBase Model=Qwen2.5-7B-Instruct2026.06 | 13.19 | — | — | 42.01 | |
| DAPOBase Model=Qwen2.5-7B-Instruct2026.06 | 13.07 | — | — | 41.8 | |
| GRPOBase Model=Qwen2.5-7B-Instruct2026.06 | 12.44 | — | — | 41.61 | |
| Math-InstructModel Scale=7B2026.06 | 12.3 | — | — | — | |
| DGPOBackbone=Qwen2.5-Math-7B2026.06 | 10.21 | — | — | — | |
| few-shotModel=OPEN-REASONER-7B2026.05 | 10 | — | — | — | |
| GRPO-ADBackbone=Qwen2.5-Math-7B2026.06 | 9.48 | — | — | — | |
| GPGBackbone=Qwen2.5-Math-7B2026.06 | 9.06 | — | — | — | |
| DAPOBackbone=Qwen2.5-Math-7B2026.06 | 8.75 | — | — | — | |
| GRPOBackbone=Qwen2.5-Math-7B2026.06 | 8.44 | — | — | — | |
| GSPOBackbone=Qwen2.5-Math-7B2026.06 | 8.33 | — | — | — | |
| Dr. GRPOBackbone=Qwen2.5-Math-7B2026.06 | 8.23 | — | — | — | |
| Math-BaseModel Scale=7B2026.06 | 7 | — | — | — | |
| ReSumBackbone=Qwen2.5-Math-1.5B2026.06 | 6.83 | — | — | — | |
| Math-InstructModel Scale=1.5B2026.06 | 6.7 | — | — | — | |
| LRSModel=OPEN-REASONER-1.5B, Prompting=0-shot2026.05 | 6.6 | — | — | — | |
| DGPOBackbone=Qwen2.5-Math-1.5B2026.06 | 5.73 | — | — | — | |
| BaseBase Model=Qwen3-4B-Base2026.06 | 5.3 | — | — | 29.4 | |
| Base ModelBackbone=Qwen2.5-Math-7B2026.06 | 4.79 | — | — | — | |
| GRPOBackbone=Qwen2.5-Math-1.5B2026.06 | 3.96 | — | — | — | |
| Base ModelBackbone=Qwen2.5-Math-1.5B2026.06 | 3.65 | — | — | — | |
| BaseBase Model=Qwen2.5-7B-Instruct2026.06 | 3.52 | — | — | 34.1 |