Python Reference-Model Generation on VerilogEval v2
82.4Pass@1ChipMate-Python-9B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ChipMate-Python-9BType=ChipMate-P, Size=9B2026.05 | 82.4 | 84.7 | |
| ChipMate-Python-4BType=ChipMate-P, Size=4B2026.05 | 77.6 | 80.1 | |
| GPT-5.5Type=Foundation Models, Size=-2026.05 | 61.5 | 75.5 | |
| DeepSeek CoderType=Foundation Models, Size=236B2026.05 | 60.1 | 73.1 | |
| DeepSeek V4Type=Foundation Models, Size=1.6T2026.05 | 59.7 | 71.8 | |
| DeepSeek R1Type=Foundation Models, Size=671B2026.05 | 57.1 | 70.7 | |
| Qwen3.5-9BType=Base Models, Size=9B2026.05 | 48 | 64.7 | |
| Qwen3.5-4BType=Base Models, Size=4B2026.05 | 46.8 | 63.7 | |
| CodeV-R1Type=Specialized Models, Size=7B2026.05 | 45.3 | 56.4 | |
| CodeV-R1 (distill)Type=Specialized Models, Size=7B2026.05 | 40.4 | 48.5 |