Research map / Reinforcement learning / Visual and representation RL
The Policy-gradient Placement and Generative Routing Neural Networks for Chip Design
Ruoyu Cheng, Xianglong Lyu, Yang Li, Junjie Ye, Jianye HAO, Junchi Yan
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 bi-discriminator to ResNet generator (ablation) | ResNet router · Route-small | CrrtR: 0.594±0.004 -> 0.738±0.005 | Improved |
| add vanilla cGAN discriminator to ResNet generator (ablation) | ResNet router · Route-small | CrrtR: 0.594±0.004 -> 0.568±0.020 | Hurt |
| add vanilla cGAN discriminator to ResNet (ablation) | ResNet router · Route-small | WLR: 1.030±0.007 -> 1.320±0.151 | Hurt |
| add enhanced loss (EL) (ablation) | bcGAN(ResNet) · Route-small-4 | WLR: 1.035±0.013 -> 1.010±0.000 | Improved |
| add enhanced loss (EL) (ablation) | bcGAN(ResNet) · Route-small | CrrtR: 0.738±0.005 -> 0.735±0.010 | No clear effect |
| replace CVAE*(CNN) backbone with ResNet plus bi-discriminator and EL (ablation) | CVAE*(CNN) · Route-small-4 | CrrtR: 0.414±0.020 -> 0.814±0.001 | Improved |
| add bi-discriminator to CVAE*(CNN) (ablation) | CVAE*(CNN) · Route-small-4 | WLR: 1.179±0.033 -> 1.525±0.029 | Hurt |
| add bi-discriminator to U-Net* (ablation) | U-Net* · Route-small-4 | WLR: 3.306±0.266 -> 1.134±0.055 | Improved |
| add generative router to RL placer (ablation) | DeepPlace · ISPD-05 adaptec1 | WL: 6149 -> 5940 | Improved |
| add net order learning (ablation) | RL placer + GR · ISPD-05 adaptec4 | RC: 121.435 -> 65.796 | Improved |
| add net order learning (ablation) | RL placer + GR · ISPD-05 adaptec3 | RC: 73.324 -> 43.207 | Improved |
| add generative router without net order learning (ablation) | DeepPlace · ISPD-05 adaptec3 | RC: 62.751 -> 73.324 | Hurt |
Closest papers by meaning
- NeuralSteiner: Learning Steiner Tree for Overflow-avoiding Global Routing in Chip Design (NeurIPS 2024 · similarity 0.90)
- Chip Placement with Diffusion Models (ICML 2025 · similarity 0.89)
- Reinforcement Learning Policy as Macro Regulator Rather than Macro Placer (NeurIPS 2024 · similarity 0.88)
- EGPlace: An Efficient Macro Placement Method via Evolutionary Search with Greedy Repositioning Guided Mutation (ICML 2025 · similarity 0.88)
- Reinforcement Learning within Tree Search for Fast Macro Placement (ICML 2024 · similarity 0.86)
- See it to Place it: Evolving Macro Placements with Vision Language Models (NeurIPS 2026 · similarity 0.86)
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.
