Tensorboard: TensorBoard feature request: Search for Node.

[chihuahua]

This issue had been migrated from tensorflow/tensorflow#2730.

I have been trying to retrain the inception v3 network and TensorBoard has been the only real way to introspect the network and it's implementation so far.

It would be very useful to be able to find nodes in the graph, by name or other properties. E.g. in my case I am interested in seeing the total_loss node that is being defined here: https://github.com/tensorflow/models/blob/master/inception/inception/inception_train.py#L120 To get a greater idea of what it is linked to.

[Wakeupbuddy]

would support the proposal of searching by name, like ctrl-F in the webpage.

[jart]

It looks like this feature request has a lot of community support. We've jotted down note to probably make this happen on Q2.

[jart]

Re: @Wakeupbuddy's note, I soft-agree, but such a path must be taken with very special caution.

1. It's almost always a good idea to seamlessly interop with the tools that already exist, since folks already understand them.

2. There's been instances in the past where this has been a suboptimal solution. For example, it is not possible for me to Ctrl+F through the comments attached to a Google Doc.

[jart]

We feel this issue should be solved. Chi won't be able to work on it this quarter. I'm marking contributions welcome, since this would be a good opportunity for someone in the community to get their feet wet. If so, please reach out to us on this issue, and we'll help guide you through the process.

[chihuahua]

As of Saturday, 4/14, users who install TensorBoard's nightly build (pip install tb-nightly) will be able to use the node search feature introduced by #1130.