Code Generation on LeetCode (OOD)
27.8AccuracyRL-PLUS
Evaluation results
| Method | Links | |
|---|---|---|
| RL-PLUSBase model=Qwen2.5-Math-7B2025.07 | 27.8 | |
| Base ModelBase model=Qwen2.5-Math-7B2025.07 | 22.8 | |
| GRPOBase model=Qwen2.5-Math-7B2025.07 | 21.1 | |
| SFT+GRPOBase model=Qwen2.5-Math-7B2025.07 | 8.34 | |
| SFTBase model=Qwen2.5-Math-7B2025.07 | 8.3 |