Mathematical Word Problems on GSM8k (OOD)
72.9AccuracyR1 Distill -> GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| R1 Distill -> GRPOtraining=SFT + GRPO2025.12 | 72.9 | |
| BOLT -> GRPOtraining=SFT + GRPO2025.12 | 69.7 | |
| STaR -> GRPOtraining=SFT + GRPO2025.12 | 68.6 | |
| SkillFactory -> GRPOtraining=SFT + GRPO2025.12 | 68.2 | |
| RL-Onlytraining=RL2025.12 | 67.7 | |
| R1 Distilltraining=SFT2025.12 | 62.9 | |
| Qwen2.5 1.5B Instructtraining=None2025.12 | 59.2 | |
| SkillFactorytraining=SFT2025.12 | 59.1 | |
| STaRtraining=SFT2025.12 | 31.1 | |
| BOLTtraining=SFT2025.12 | 23.4 |