Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/cornell-zhang/hoga

Hop-Wise Graph Attention for Scalable and Generalizable Learning on Circuits

https://github.com/cornell-zhang/hoga

Science Score: 23.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ○
    .zenodo.json file
  • ○
    DOI references
  • ✓
    Academic publication links
    Links to: arxiv.org, zenodo.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (8.7%) to scientific vocabulary

Keywords

distributed-training electronic-design-automation graph-learning graph-transformer logic-synthesis
Last synced: 5 months ago · JSON representation

Repository

Hop-Wise Graph Attention for Scalable and Generalizable Learning on Circuits

Basic Info
  • Host: GitHub
  • Owner: cornell-zhang
  • License: bsd-3-clause
  • Language: Python
  • Default Branch: master
  • Homepage:
  • Size: 635 KB
Statistics
  • Stars: 21
  • Watchers: 8
  • Forks: 2
  • Open Issues: 1
  • Releases: 0
Topics
distributed-training electronic-design-automation graph-learning graph-transformer logic-synthesis
Created almost 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme License

README.md

HOGA

arXiv

HOGA is an attention model for scalable and generalizable learning on circuits. By leveraging a novel gated attention module on hop-wise features, HOGA not only outperforms prior graph learning models on challenging circuit problems, but is also friendly to distributed training by mitigating communication overhead caused by graph dependencies. This renders HOGA applicable to industrial-scale circuit applications. More details are available in our paper.

| HOGA.png | |:--:| | Figure1: An overview of HOGA and gated attention module. |

Requirements

  • python 3.9
  • pytorch 1.12 (CUDA 11.3)
  • torch_geometric 2.1

Datasets

Pre-processed CSA and Booth Multipliers (for Gamora experiments)

Check at: https://huggingface.co/datasets/yucx0626/Gamora-CSA-Multiplier/tree/main

Pre-processed OpenABC-D benchmark (for OpenABC-D experiments)

Check at: https://zenodo.org/records/6399454#.YkTglzwpA5k

Note

The implementation of hop-wise feature generation is available here. The model (i.e., hop-wise gated attention) implementation is available here. You can adjust them for your own tasks.

Citation

If you use HOGA in your research, please cite our work published in DAC'24.

@inproceedings{deng2024hoga, title={Less is More: Hop-Wise Graph Attention for Scalable and Generalizable Learning on Circuits}, author={Chenhui Deng and Zichao Yue and Cunxi Yu and Gokce Sarar and Ryan Carey and Rajeev Jain and Zhiru Zhang}, booktitle={DAC}, year={2024}, }

Owner

  • Name: Cornell Zhang Research Group
  • Login: cornell-zhang
  • Kind: organization

GitHub Events

Total
  • Watch event: 9
  • Fork event: 2
Last Year
  • Watch event: 9
  • Fork event: 2