Research map / Vision-language and multimodal / Human and embodied modeling
CircuitNet 3.0: A Multi-Modal Dataset with Task-Oriented Augmentation for AI-Driven Circuit Design
Mingjun Wang, Yihan Wen, Yuntao Lu, Fengrui Liu, Yuxiang Zhao, Boyu Han, Jianan Mu, Yibo Lin, Runsheng Wang, Bei Yu, Huawei Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add multi-modal (RTL plus netlist) learning (ablation) | MOSS · Netlist power prediction (Toggle Rate) | PCC: 0.674 -> 0.871 | Improved |
| add multi-modal (RTL plus netlist) learning (ablation) | MOSS · Netlist power prediction (Total Power) | PCC: 0.815 -> 0.948 | Improved |
| add multi-modal (RTL plus netlist) learning (ablation) | MOSS · Netlist power prediction (Total Power) | MAPE: 27.7% -> 7.4% | Improved |
| add cross-stage knowledge distillation from layout (RTLDistil vs RTL-Timer) | RTL timing predictor · RTL timing prediction (Arrival Time) | PCC: 0.835 -> 0.887 | Improved |
| add cross-stage knowledge distillation from layout (RTLDistil vs RTL-Timer) | RTL timing predictor · RTL timing prediction (TNS) | PCC: 0.801 -> 0.918 | Improved |
| train on CircuitNet 3.0 augmented data instead of Resyn-27k | RTLDistil · RTL timing prediction (Arrival Time) | PCC: 0.842 -> 0.935 | Improved |
| train on CircuitNet 3.0 augmented data instead of Resyn-27k | RTLDistil · RTL timing prediction (Arrival Time) | MAPE: 23.86% -> 15.28% | Improved |
| add task-oriented augmentation to original data (ablation) | RTLDistil · RTL timing prediction (Arrival Time) | PCC: 0.887 -> 0.935 | Improved |
| add task-oriented augmentation to original data (ablation) | MasterRTL · RTL timing prediction (Arrival Time) | PCC: 0.520 -> 0.613 | Improved |
| add task-oriented augmentation to original data (ablation) | RTLTimer · RTL timing prediction (Arrival Time) | PCC: 0.835 -> 0.873 | Improved |
| train on augmented data instead of original data (ablation) | VIRTUAL · RTL total power prediction | PCC: 0.672 -> 0.753 | Improved |
| train on augmented data instead of Resyn-27k | VIRTUAL · RTL total power prediction | MAPE: 27.48% -> 23.92% | Improved |
Closest papers by meaning
- CircuitFusion: Multimodal Circuit Representation Learning for Agile Chip Design (ICLR 2025 · similarity 0.92)
- DeepGate4: Efficient and Effective Representation Learning for Circuit Design at Scale (ICLR 2025 · similarity 0.89)
- AnalogGenie: A Generative Engine for Automatic Discovery of Analog Circuit Topologies (ICLR 2025 · similarity 0.89)
- DeepLayout: Learning Neural Representations of Circuit Placement Layout (ICML 2025 · similarity 0.88)
- FALCON: An ML Framework for Fully Automated Layout-Constrained Analog Circuit Design (NeurIPS 2025 · similarity 0.88)
- Circuit Representation Learning with Masked Gate Modeling and Verilog-AIG Alignment (ICLR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
