Code Generation on HumanEval+ (Score (%))
51.2Score (%)Qwen3-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-4BParameters=4B2026.05 | 51.2 | |
| Qwen2.5-7BParameters=7B2026.05 | 47 | |
| Qwen3.5-4BParameters=4B2026.05 | 43.9 | |
| OLMo-3-7BParameters=7B2026.05 | 39.6 | |
| Mellum 2Parameters=2.5B/12B2026.05 | 37.2 |