Mathematical Reasoning on GSM Hard (Accuracy, Micro avg.)
52.69AccuracyTraceLift
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TraceLiftModel=Qwen2.5-14B2026.05 | 52.69 | 72.25 | |
| Exec-onlyModel=Qwen2.5-14B2026.05 | 52.39 | 71.93 | |
| TraceLiftModel=Qwen2.5-7B2026.05 | 52.16 | 70.19 | |
| BaseModel=Qwen2.5-14B2026.05 | 52.16 | 71.2 | |
| Exec-onlyModel=Qwen2.5-7B2026.05 | 47.92 | 65.62 | |
| BaseModel=Qwen2.5-7B2026.05 | 30.02 | 46.51 | |
| TraceLiftModel=Llama3.1-8B2026.05 | 10 | 22.86 | |
| Exec-onlyModel=Llama3.1-8B2026.05 | 9.78 | 22.19 | |
| BaseModel=Llama3.1-8B2026.05 | 4.93 | 12.74 |