Mathematical Reasoning on Countdown-4
60.77AccuracyDS (Oracle)
Evaluation Results
| Method | Links | |
|---|---|---|
| DS (Oracle)Backbone=Qwen2.5-7B, Rollouts=1006k, Finetuned=true2026.03 | 60.77 | |
| DPSBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.03 | 59.61 | |
| HRBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.03 | 54.54 | |
| USBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.03 | 53.27 | |
| DPSBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.03 | 47.78 | |
| DS (Oracle)Backbone=Qwen2.5-3B, Rollouts=1141k, Finetuned=true2026.03 | 47.67 | |
| HRBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.03 | 42.1 | |
| USBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.03 | 39.42 |