Matthew Rocklin

Noah, have you taken a look at the caching that Dask already provides? http://dask.pydata.org/en/latest/caching.html On Wed, Mar 22, 2017 at 2:29 PM, Noah D Brenowitz wrote: > I just wanted...

Perhaps with an on-disk mutablemapping like shelve or chest. Regardless, I don't think that this is related to Zarr, so we should probably move the discussion to a different issue....

That's great. When I started looking into this I quickly became lost on how to set up the problem. You appear to have enough practical experience that that's not much...

It depends on what you mean by "batch" I guess. You can slice into x in a variety of ways ```python index = np.random.randint(0, x.shape[0], size=10) batch = x[index] ```...

The same as you would with NumPy ```python for i in range(0, x.shape[0], 32): chunk = x[i:i+32, ...] ``` chunk is a dask array here. I'm not sure if that's...

This keeps coming up. I'm adding it to the core maintenance project board. https://stackoverflow.com/questions/61924824/how-to-do-model-predict-using-distributed-dask-with-a-pre-trained-keras-model

(small note, we've been using the term "name" for this with other Dask classes, notably workers and clients) On Mon, Oct 7, 2019 at 5:05 AM Eric Dill wrote: >...

My understanding of @ian-r-rose 's original plan was that Dask-Gateway itself might replace the server-side component in dask-labextension. That way JupyterLab might talk directly to routes on the Gateway server.

It looks like there is an example in the documentation here: http://dask-ml.readthedocs.io/en/latest/examples/xgboost.html It's nice in many respects (real data, easily interpretable problem, ...) However a couple things are concerning about...

Sounds fun On Sun, Jul 12, 2020 at 3:32 PM Scott Sievert wrote: > The paper "BOHB: Robust and Efficient Hyperparameter Optimization at Scale > " includes an > interesting...