IBM Watson Data & AI - Structured Ideas

Welcome to the idea forum for IBM Watson Data & AI — our team welcomes any feedback, requests, and suggestions you have for improving our products! 

This forum allows us to connect your product improvement ideas with IBM product and engineering teams. 

Provide a run button on Visual NN flow editor to simplify training and testing the network while prototyping

As a deep learning novice user, when I am prototyping, I want to be able to easily run a network so that I can try different configurations without having to define an Experiment.

The Deep Learning with Python book encourages you to try different network configurations while learning Keras.  After describing each neural network, it encourages trying out different things, e.g.

  • Try using larger or smaller layers: 32 units, 128 units, and so on.

  • You used two hidden layers. Now try using a single hidden layer, or three hid-

    den layers.

It would be good if users could easily try out things like this without having to deploy the training definition and create an Experiment.

The current editor assumes you will not be iterating your design much when creating a network.

  • Chris Snow
  • May 1 2018
  • Needs review
Role Summary
  • Attach files
  • Chris Snow commented
    May 1, 2018 09:53
  • Atsushi Tsuchiya commented
    May 24, 2018 04:04

    I am getting some feedback from those who already uses GPU says that they wanna keep existing framework - that is Jupyter/Tensorflow to GPU - not propriety mechanism like Experiments - They are not saying the Experiments functionality is not good but simply they wanna have both way of executing Tensorflow on GPUs.