Post-synthesis Area Prediction on RTL Design Dataset (val)
0.2982MAEPM-based Predictor
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PM-based PredictorKnowledge Distillation=true, Role=Teacher2025.08 | 0.2982 | 5 | 0.9334 | 0.2581 | |
| StructRTLKnowledge Distillation=true2025.08 | 0.3856 | 7 | 0.8676 | 0.3639 | |
| StructRTL (w/o Lmnm)Knowledge Distillation=true, Ablation=without masked node modeling loss2025.08 | 0.4015 | 7 | 0.8557 | 0.3799 | |
| StructRTL (w/o Lep)Knowledge Distillation=true, Ablation=without edge prediction loss2025.08 | 0.4071 | 7 | 0.848 | 0.3899 | |
| CodeV-CL-7BKnowledge Distillation=true, Category=LLM-Based2025.08 | 0.4192 | 8 | 0.8225 | 0.4213 | |
| CodeV-QW-7BKnowledge Distillation=true, Category=LLM-Based2025.08 | 0.4397 | 8 | 0.8174 | 0.4273 | |
| Graph-GNNKnowledge Distillation=true, Category=Graph-Based2025.08 | 0.4689 | 9 | 0.7954 | 0.4523 | |
| CodeV-DS-6.7BKnowledge Distillation=true, Category=LLM-Based2025.08 | 0.4896 | 9 | 0.7928 | 0.4552 |