Code Generation on HumanEval compile (L1)
0.1847Pass@1ShieldedCode
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ShieldedCode2026.01 | 0.1847 | 0.2794 | |
| GPT-4o2026.01 | 0.1743 | 0.2518 | |
| DeepSeekCoder-7BParameters=7B2026.01 | 0.0689 | 0.1065 | |
| GPT-3.5-Turbo2026.01 | 0.0571 | 0.0795 | |
| Meta LLMCompiler-7BParameters=7B2026.01 | 0.0538 | 0.0689 | |
| Qwen-2.5-Coder-7BParameters=7B2026.01 | 0.0512 | 0.0728 | |
| StarCoder2-7BParameters=7B2026.01 | 0.0491 | 0.0473 | |
| CodeLlama2026.01 | 0.0326 | 0.0534 |