Code Generation on HumanEval (Score %)
57.3Score (%)Qwen3-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-4BParameters=4B2026.05 | 57.3 | |
| Qwen2.5-7BParameters=7B2026.05 | 55.5 | |
| Qwen3.5-4BParameters=4B2026.05 | 50 | |
| OLMo-3-7BParameters=7B2026.05 | 45.1 | |
| Mellum 2Parameters=2.5B/12B2026.05 | 41.5 |