Mathematical Reasoning on GSM8K (test) (0-shot and 5-shot Accuracy)
44.8Accuracy (5-shot)CRESCENT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CRESCENTTeacher data=GSM8K, #Data=50k, Teacher model=Llama3-8B-I.2025.02 | 44.8 | 30.8 | |
| MetaMathTeacher data=GSM8K, #Data=50k, Teacher model=Llama3-8B-I.2025.02 | 41.7 | 22 | |
| ScaleQuestTeacher data=GSM8K&MATH, #Data=50k, Teacher model=Mix2025.02 | 38.9 | 22.8 | |
| GSM8K BaselineTeacher data=GSM8K, #Data=7k2025.02 | 38.4 | 38.4 | |
| MMIQCTeacher data=Mix, #Data=50k, Teacher model=GPT-42025.02 | 33.7 | 28.3 |