Skip to content
This repository has been archived by the owner on May 28, 2024. It is now read-only.

Latest commit

 

History

History
29 lines (25 loc) · 1.36 KB

README.md

File metadata and controls

29 lines (25 loc) · 1.36 KB

tfjs-mnist-workshop

This is an e2e TensorFlow workshop from model training using Keras API all the way to visualization using TensorFlow.js

User will need to complete exercises (.ipynb files) under py/ folder. Reference implementation could be found from /py/solutions folder.

User guide

  • Prepare environment
    • For Windows users, download and install ANaconda with either Python 2 or Python 3
    • For Linux/Mac users, you should already have Python installed
  • Install dependencies
    • pip install -r requirements.txt
    • If you're facing connectivity issues, please use pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt
  • Visualize the initial model in browser
    • For Python 2 users, python -m SimpleHTTPServer
    • For Python 3 users, python -m http.server
    • Open localhost:8080 in your browser
    • Most of the predictions are wrong (red) because the model is doing random guess
  • Coding
    • Use jupyter notebook to complete an exercise and generate a model
  • Convert the mdoel to TF JS format
    • For windows users, convert.bat PATH_TO_THE_H5_FILE
    • For Linux/Mac users, convert.sh PATH_TO_THE_H5_FILE
  • Visualize the model in browser
    • For Python 2 users, python -m SimpleHTTPServer
    • For Python 3 users, python -m http.server
    • Open localhost:8080 in your browser
    • Most of the predictions should be correct (green) now