Mathematical Reasoning on Minerva (Accuracy, Avg.)
45.92AccuracySTRIDE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| STRIDEBase Model=Qwen2.5-7B-Instruct2026.06 | 45.92 | 44.71 | |
| STRIDEBase Model=Qwen3-4B-Base2026.06 | 43.28 | 49.24 | |
| HET-DAPOBase Model=Qwen2.5-7B-Instruct2026.06 | 42.97 | 42.18 | |
| HICRA-GRPOBase Model=Qwen2.5-7B-Instruct2026.06 | 42.42 | 42.01 | |
| DAPOBase Model=Qwen2.5-7B-Instruct2026.06 | 42.21 | 41.8 | |
| GRPOBase Model=Qwen2.5-7B-Instruct2026.06 | 41.82 | 41.61 | |
| HET-DAPOBase Model=Qwen3-4B-Base2026.06 | 40.69 | 47.27 | |
| DAPOBase Model=Qwen3-4B-Base2026.06 | 40.67 | 46.63 | |
| HICRA-GRPOBase Model=Qwen3-4B-Base2026.06 | 40.23 | 47.09 | |
| BaseBase Model=Qwen2.5-7B-Instruct2026.06 | 39.16 | 34.1 | |
| GRPOBase Model=Qwen3-4B-Base2026.06 | 38.65 | 45.28 | |
| STRIDEBase Model=Llama-3.1-8B-Instruct2026.06 | 28.82 | 25.87 | |
| BaseBase Model=Qwen3-4B-Base2026.06 | 28.3 | 29.4 | |
| HET-DAPOBase Model=Llama-3.1-8B-Instruct2026.06 | 26.47 | 23.97 | |
| GRPOBase Model=Llama-3.1-8B-Instruct2026.06 | 26.4 | 22.92 | |
| DAPOBase Model=Llama-3.1-8B-Instruct2026.06 | 26.05 | 23.41 | |
| HICRA-GRPOBase Model=Llama-3.1-8B-Instruct2026.06 | 25.93 | 23.3 | |
| BaseBase Model=Llama-3.1-8B-Instruct2026.06 | 20.9 | 17.78 |