Code Generation on LiveCodeBench 2025.02.01-2025.05.01 v6
38.3Average Score (@16)ASPO-Code-4B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ASPO-Code-4BBase Model=Qwen3-4B2025.10 | 38.3 | 47.3 | 41.6 | |
| Archer-Code-4B2025.07 | 37.5 | 48.1 | 39.8 | |
| Qwen3-4B (GRPO)training_algorithm=GRPO2025.07 | 36.6 | 45 | 38.7 | |
| GRPOBase Model=Qwen3-4B2025.10 | 36.6 | 45 | 38.7 | |
| ASPO-Code-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 30.5 | 46 | 31 | |
| Archer-Code-1.5B2025.07 | 30.2 | 45.8 | 29.8 | |
| Nemotron-1.5B2025.07 | 29.5 | 42.8 | 27.8 | |
| Nemotron-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 29.5 | 42.8 | 27.8 | |
| DeepSeek-R1-1.5B (GRPO)training_algorithm=GRPO2025.07 | 27.6 | 43.5 | 26.8 | |
| GRPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 27.6 | 43.5 | 26.8 | |
| Qwen3-4B2025.07 | 24 | 35.1 | 23.9 | |
| Qwen3-4BBase Model=Qwen3-4B2025.10 | 24 | 35.1 | 23.9 | |
| DeepCoder-1.5B2025.07 | 22.6 | 42 | 23 | |
| DeepCoder-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 22.6 | 42 | 23 | |
| DeepSeek-R1-1.5B2025.07 | 17.2 | 34.4 | 17 | |
| DeepSeek-R1-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 17.2 | 34.4 | 17 |