Overview
SUAVE is an external web app run by Ilya Zaslavsky at suave.sdsc.edu. It can be used with integration of notebooks between Datahub and applications on the SuAVE server. This page is to address some issues with accessing the SuAVE tool through Datahub.
Accessing SUAVE on Datahub
Note: The cloning instructions should be done from a Datahub or dsmlp-login terminal, not a git, bash, or local terminal. For more information on this, visit the article "How to: Launching Containers From the Command Line - Data Science/Machine Learning Platform (DSMLP)".
Sources:
For students who have used SUAVE in a current course:
- Run: “git clone –depth 1 https://github.com/suave-ucsd/jupyter-suave.git”
For students who no longer have access SUAVE through a course:
- Clone Suave GitHub repo to your datahub: “git clone –depth 1 https://github.com/suave-ucsd/jupyter-suave.git”.
- Create a SuAVE account at suave-dev.sdsc.edu, and login.
- Navigate to the galleries you are interested in (e.g. Margie Burton’s account). Margie is an archaeologist. You can use a clone of one of them using the gallery urls
- e.g. "http://suave-dev.sdsc.edu/main/file=zaslavsk_Wadi_Fidan_61_Pottery_Neolithic_clone_.csv&views=1110101&view=grid" (For this one full-size images are available. Full-size images for other collections are in the process of being published.).
- Click the Jupyter button in SuAVE (top right corner).
- Click Clone Survey: a copy of this survey will be created in your survey gallery.
- Edit the survey in your gallery, and make sure Jupyter is enabled.
- When you view your survey, click the Jupyter button, and select Datahub as the hub – then try to run some existing notebooks to see how it works. Select one of existing notebooks that can access images.
- Modify your transfer learning notebook to be similar to other SuAVE notebooks (that is, it will have several cells at the beginning, to accept parameters, and a couple cells at the end), to publish results of the modeling back into Suave.