Douglas Ashton

One group changed the units to millions $ (/1e6) which kept the context but had the benefit of scaling.

I would say class imbalance is the same in keras as with any machine learning problem. So you can down/up sample outside of keras (potentially with something like `{rsample}`) and...

Nearest I could find in google https://imbalanced-learn.readthedocs.io/en/stable/generated/imblearn.keras.BalancedBatchGenerator.html but it would need some reticulating for R.

Only scanned the abstracts there but if time ordering is important then bringing in an RNN or LSTM will allow for sequences that are time ordered.

I can reproduce this problem. Investigating. I think I'm running out of GPU memory. Which wasn't a problem before. I do have two sessions running but not sure if that's...

Dropping the batch size to 5 has got it moving again.

Not sure if this helps but I've had some success using the [tensorflow-macos](https://pypi.org/project/tensorflow-macos/) library to get arm based wheels. It's a fork by Apple that supports M1. Inside poetry we...