Mathematical Reasoning on AQuA (FRS and Pass@1 Accuracy)
96.8FRSDS-R1-7B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DS-R1-7BFiltering rate=10%, Decoding strategy=greedy2026.04 | 96.8 | — | |
| DS-R1-1.5BFiltering rate=10%, Decoding strategy=greedy2026.04 | 90.1 | — | |
| Phi-4-Reas.Filtering rate=10%, Decoding strategy=greedy2026.04 | 83.2 | — | |
| Qwen3-4BFiltering rate=10%, Decoding strategy=greedy2026.04 | 78.5 | — | |
| Phi-4Filtering rate=10%, Decoding strategy=greedy2026.04 | 78.5 | — | |
| Qwen2.5-7BFiltering rate=10%, Decoding strategy=greedy2026.04 | 77.2 | — | |
| Qwen2.5-MathFiltering rate=10%, Decoding strategy=greedy2026.04 | 74.9 | — | |
| LLaMA-3.1-8BFiltering rate=10%, Decoding strategy=greedy2026.04 | 52.9 | — | |
| Gemma-7BFiltering rate=10%, Decoding strategy=greedy2026.04 | 22.2 | — | |
| COLMBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 58.27 | |
| GradNormBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 59.45 | |
| GREATSBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 61.42 | |
| IDBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 58.66 | |
| IWDBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 59.8 | |
| PartitionSelBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 62.2 | |
| RandomBackbone=Qwen2.5-3B, Fine-tuning Dataset=Meta-MathQA, Subset Fraction=50%2026.06 | — | 60.86 |