Travis is not going to automatically offer the free tier for all open source projects; We likely want o migrate away from travis.

Setting up github actions to replace travis would be a welcomed contribution.

/kind feature

This is a sub-issue from #5494.
We need to remove any f"strings" from the backend of the Jupyter web app in order to be able to use lower versions from Python 3.6 to run it.

/good-first-issue

Current pytorch implementation ignores the argument split_f in the function train_batch_ch13 as shown below.

def train_batch_ch13(net, X, y, loss, trainer, devices):
    if isinstance(X, list):
        # Required for BERT Fine-tuning (to be covered later)
        X = [x.to(devices[0]) for x in X]
    else:
        X = X.to(devices[0])
...

Todo: Define the argument `

Hello,

As suggested in #823 by @ABCurado it could be interesting to talk about the build history and to put the link to the wiki in

• The README.md,
• The documentation.

Best.

The plotting UI leaves a lot to be desired:

• When you bring up the plotting UI, it's not obvious that you have to do some work to make a plot happen. If you don't, there's no error message, just a never-ending wait. (see https://gitter.im/polynote/polynote?at=5e0e36f9eac8d1511e9ed2ff )
• Why do we make you drag things onto axes? Especially when there is only one axis it could go onto (at leas

Use case

Polyaxon tracking can be used to automatically log all information generated by Ludwig.

We should probably think about adding support for Polyaxon's tracking module in Ludwig contrib.

I wrote this as an experiment, anyone should feel free to continue working on it! The first step is to try it out and make a list of TODOs.

https://user-images.githubusercontent.com/6933510/110047855-e9e5ba80-7d4e-11eb-9c25-ae3cf8f56b13.mov