Lee Newberg

> if the covariance matrix is a negative scalar (e.g. -2), the determinant should be negative, and it is always non-negative. I'm not sure what is meant by a "covariance...

I use `torch.utils.tensorboard.SummaryWriter` and TensorBoard. Would that do what you want? If not, what are you looking for?

I agree with @jamesobutler that we would benefit from tighter integration between the sets of checks. If it isn't already, perhaps you could add that as an issue. Should we...

This latest push adds a check that both a space and a capital letter follow the colon. It also simplifies the `if` statement to use `egrep`.

Note that the checks that the line length is at least 8 and at most 78 occur earlier in the code, but could be added here with ```bash if echo...

I like the approach of creating and using `ITK_EXERCISE_BASIC_OBJECT_METHODS_SELF`. If some day we figure out how to merge it into `ITK_EXERCISE_BASIC_OBJECT_METHODS` then we can quickly find and modify all the...

If we decide to look at using a database then there is the consideration of server side vs. client side. That is, if the goal is along the lines of...

@cooperlab says: look at TF MultiWorker strategy - https://www.tensorflow.org/tutorials/distribute/multi_worker_with_keras. We can help with this. Key questions are: - Can multiple workers run on 1 machine (example suggests so) - Can...

Tensorflow does autosharding so we shouldn't have to explicitly shard the `tensorflow.Dataset`. We could add convenience functions so that the the likes of `global_batch_size = num_workers * batch_size_per_worker` are satisfied....

@thewtex Sphinx offers quite a bit of functionality and I am unsure how to prioritize. Can you recommend a project similar enough to HistomicsStream in scope that is already using...