Skip to content
Louise Lessél edited this page May 8, 2019 · 6 revisions

Homework:

  • Watch these Videos:

  • Coding:

    • Work on your final project, add you experiment link or blog below.
    • Or Train your own style transfer model and run the model in the browser with ml5.js
    • Or use Runway's Adaptive-Style-Transfer, Arbitrary-Image-Stylization model to send images to your p5 sketch
    • Or train any other models (LSTM, Word2Vec), run the model in the browser.
    • Publish your project on github or your own blog, or record a video and put it on your blog. Add your project link below.

Some issues you might have when training the model with spell

  • During step 1 preparing your environment, after you add your style image to the style folder, you need to commit the changes to git, so the remote spell machine can get access to your changes.

    $ git add images ckpt
    $ git commit -m "Added required folders and images"
    
  • When you are on step 3 Training with style.py, you need to choose a GPU machine type, CPU machine wouldn't work.

    --machine-type V100 OR --machine-type K80
    
  • On step 4 Converting model to ml5js, you need to install tensorflow on your local computer. See more instruction here. Or you can run these command on spell, so you don't need to install tensorflow locally, after you are done, spell cp your model back to your local computer.

See demos live

Resource:

Questions?

  • add your question here

Add your name / link to your assignment below

Clone this wiki locally