Mathematical Reasoning on Minerva-Math (maj@16, pass@16)
46.32maj@16PPC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PPCBackbone=Qwen3-4B2026.05 | 46.32 | 60.66 | |
| PTA-GRPOBackbone=Qwen3-4B2026.05 | 45.22 | 58.09 | |
| Plan-TuningBackbone=Qwen3-4B2026.05 | 44.85 | 56.25 | |
| PPCBackbone=Qwen2.5-7B2026.05 | 44.85 | 62.87 | |
| GRPOBackbone=Qwen3-4B2026.05 | 44.48 | 55.88 | |
| Prompt-OnlyBackbone=Qwen3-4B2026.05 | 43.75 | 54.41 | |
| PPCBackbone=Qwen2.5-Math-7B2026.05 | 43.38 | 58.09 | |
| BaseBackbone=Qwen3-4B2026.05 | 43.01 | 55.15 | |
| PTA-GRPOBackbone=Qwen2.5-7B2026.05 | 43.01 | 58.82 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.05 | 42.65 | 53.31 | |
| Prompt-OnlyBackbone=Qwen2.5-Math-7B2026.05 | 41.92 | 52.21 | |
| Plan-TuningBackbone=Qwen2.5-7B2026.05 | 41.54 | 55.88 | |
| GRPOBackbone=Qwen2.5-7B2026.05 | 41.18 | 61.76 | |
| BaseBackbone=Qwen2.5-Math-7B2026.05 | 40.81 | 51.84 | |
| BaseBackbone=Qwen2.5-7B2026.05 | 40.07 | 54.78 | |
| Plan-TuningBackbone=Qwen2.5-Math-7B2026.05 | 40.07 | 52.94 | |
| Prompt-OnlyBackbone=Qwen2.5-7B2026.05 | 39.71 | 58.09 | |
| PTA-GRPOBackbone=Qwen2.5-Math-7B2026.05 | 37.87 | 56.62 | |
| PPCBackbone=Llama3.1-8B2026.05 | 37.5 | 61.4 | |
| PTA-GRPOBackbone=Llama3.1-8B2026.05 | 37.13 | 59.93 | |
| GRPOBackbone=Llama3.1-8B2026.05 | 36.4 | 59.19 | |
| Plan-TuningBackbone=Llama3.1-8B2026.05 | 36.03 | 58.09 | |
| BaseBackbone=Llama3.1-8B2026.05 | 35.29 | 60.29 | |
| Prompt-OnlyBackbone=Llama3.1-8B2026.05 | 34.56 | 59.56 |