Pytorx

<!-- --- markdown: image_dir: /assets path: README.md ignore_from_front_matter: true absolute_image_path: false #是否使用绝对（相对于项目文件夹）图片路径 --- --> <p align="center"> <img src="./imgs/pytorx_logo3.jpeg" alt="PytorX: s" width="850"> <br> <!-- <a href="https://travis-ci.org/lord/slate"><img src="https://travis-ci.org/lord/slate.svg?branch=master" alt="Build Status"></a> --> </p> <p align="center">PytorX helps you evaluate Neural Network performance on Crossbar Accelerator.</p>

Features

• This is the alpha version of PytorX, a beta version will be released shortly
• Clean and Easy-to-Ues <!-- — Built on pytorch and GPU enabled -->
• Evaluation for Research of Device/Circuit/Architecture

<!-- * **Monitor integrated** — The functions -->

Getting Started with PytorX

This project aims at building an easy-to-use framework for neural network mapping on crossbar-based accelerator with resistive memory (e.g., ReRAM, MRAM, etc.).

• Dependencies
• Usage
• Results
• Methods

If you find this project useful to you, please cite our work:

@inproceedings{He2019NIA,
  title={Noise Injection Adaption: End-to-End ReRAM Crossbar Non-ideal Effect Adaption for Neural Network Mapping},
  author={He, Zhezhi and Lin, Jie and Ewetz, Rickard and Yuan, Jiann-Shiun and Fan, Deliang},
  booktitle={Proceedings of the 56th Annual Design Automation Conference},
  pages={105},
  year={2019},
  organization={ACM}
}

Dependencies:

• Python 3.6 (Anaconda)
• Pytorch 1.1
• cuDNN

Python package installation

Set the environment variable PYTHONPATH to locate the library. For example, assume we cloned pytorch repository on the home directory ~. then we can added the following line in ~/.bashrc.

export PYTORX_HOME=/path/to/pytorx
export PYTHONPATH=$PYTORX_HOME/python:${PYTHONPATH}

sample code on author's machine:

export PYTORX_HOME=/Users/elliot/Dropbox/Github/PytorX
export PYTHONPATH=$PYTORX_HOME/python:${PYTHONPATH}

Then you are ready to go~

Usage

Simply run

$ bash run.sh

to execute a MNIST example.