Code Generation on CodeAlpaca 20k (NLL)
0.597NLLSPD
Evaluation Results
| Method | Links | |
|---|---|---|
| SPDBackbone=Qwen2.5-7B-Instruct2026.05 | 0.597 | |
| PSRBackbone=Qwen2.5-7B-Instruct2026.05 | 0.603 | |
| SSDBackbone=Qwen2.5-7B-Instruct2026.05 | 0.608 | |
| Base modelBackbone=Qwen2.5-7B-Instruct2026.05 | 0.653 | |
| SPDBackbone=Qwen2.5-0.5B-Instruct2026.05 | 0.679 | |
| SSDBackbone=Qwen2.5-0.5B-Instruct2026.05 | 0.682 | |
| Base modelBackbone=Qwen2.5-0.5B-Instruct2026.05 | 0.683 | |
| PSRBackbone=Qwen2.5-0.5B-Instruct2026.05 | 0.683 | |
| SPDBackbone=Qwen2.5-14B-Instruct2026.05 | 0.687 | |
| SSDBackbone=Qwen2.5-14B-Instruct2026.05 | 0.688 | |
| PSRBackbone=Qwen2.5-14B-Instruct2026.05 | 0.692 | |
| SPDBackbone=Llama-3.1-8B-Instruct2026.05 | 0.692 | |
| SSDBackbone=Llama-3.1-8B-Instruct2026.05 | 0.729 | |
| Base modelBackbone=Qwen2.5-14B-Instruct2026.05 | 0.737 | |
| SPDBackbone=Qwen3-4B-Instruct2026.05 | 0.745 | |
| Base modelBackbone=Llama-3.1-8B-Instruct2026.05 | 0.751 | |
| PSRBackbone=Qwen3-4B-Instruct2026.05 | 0.752 | |
| SSDBackbone=Qwen3-4B-Instruct2026.05 | 0.754 | |
| PSRBackbone=Llama-3.1-8B-Instruct2026.05 | 0.786 | |
| Base modelBackbone=Qwen3-4B-Instruct2026.05 | 0.809 |