Mathematical Reasoning on OlympiadBench (maj@16, pass@16)
67.03Maj@16PPC
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PPCBackbone=Qwen3-4B2026.05 | 67.03 | 78.35 | |
| Prompt-OnlyBackbone=Qwen3-4B2026.05 | 66.48 | 77.25 | |
| BaseBackbone=Qwen3-4B2026.05 | 66.04 | 76.81 | |
| Plan-TuningBackbone=Qwen3-4B2026.05 | 66.04 | 77.25 | |
| GRPOBackbone=Qwen3-4B2026.05 | 64.62 | 77.91 | |
| PTA-GRPOBackbone=Qwen3-4B2026.05 | 59.89 | 76.15 | |
| PPCBackbone=Qwen2.5-Math-7B2026.05 | 46.04 | 66.26 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.05 | 45.05 | 64.18 | |
| PPCBackbone=Qwen2.5-7B2026.05 | 44.73 | 63.85 | |
| Prompt-OnlyBackbone=Qwen2.5-Math-7B2026.05 | 44.62 | 64.51 | |
| Plan-TuningBackbone=Qwen2.5-Math-7B2026.05 | 43.95 | 64.73 | |
| PTA-GRPOBackbone=Qwen2.5-Math-7B2026.05 | 43.52 | 61.54 | |
| BaseBackbone=Qwen2.5-Math-7B2026.05 | 43.41 | 53.96 | |
| GRPOBackbone=Qwen2.5-7B2026.05 | 43.19 | 59.56 | |
| Plan-TuningBackbone=Qwen2.5-7B2026.05 | 41.53 | 58.13 | |
| Prompt-OnlyBackbone=Qwen2.5-7B2026.05 | 40.77 | 61.32 | |
| PTA-GRPOBackbone=Qwen2.5-7B2026.05 | 40.77 | 62.09 | |
| BaseBackbone=Qwen2.5-7B2026.05 | 40.11 | 57.47 | |
| PPCBackbone=Llama3.1-8B2026.05 | 27.8 | 51.43 | |
| GRPOBackbone=Llama3.1-8B2026.05 | 25.05 | 42.2 | |
| PTA-GRPOBackbone=Llama3.1-8B2026.05 | 24.4 | 47.47 | |
| Prompt-OnlyBackbone=Llama3.1-8B2026.05 | 23.08 | 42.42 | |
| Plan-TuningBackbone=Llama3.1-8B2026.05 | 22.64 | 47.14 | |
| BaseBackbone=Llama3.1-8B2026.05 | 21.1 | 41.32 |