dask-ml

[westurner]

Dask-ml

Dask-ML enables parallel and distributed machine learning using Dask alongside existing machine learning libraries like Scikit-Learn.

• Src: https://github.com/dask/dask-ml
• Docs: https://dask-ml.readthedocs.io/en/latest/

...

• XGBoost, TensorFlow, scikit-learn

[nerdcha]

Looks interesting! Am I right in thinking that it adds value on clusters, rather than multicore single machines? (We only support single VMs right now, but will add clusters down the track.)

[westurner]

On Thursday, November 16, 2017, Jamie Hall ***@***.***> wrote: Looks interesting! Am I right in thinking that it adds value on clusters, rather than multicore single machines?

dask.distributed is useful for clusters. dask is generally useful for data larger than available RAM; which, given the recent Kaggle dataset max size increase to 10gb, will be more and more common.

(We only support single VMs right now, but will add clusters down the track.)

This image may be already be especially useful for multi-machine computation as well.

[nerdcha]

Sorry for the delay in getting back to you. I'll add dask-ml and its xgboost extension to the image shortly. Looks like they also have an early version of a TensorFlow interface, but its pip installation is failing, so we'll have to come back to that.

[westurner]

Thanks!

…

On Tue, Nov 28, 2017 at 8:35 PM, Jamie Hall ***@***.***> wrote: Sorry for the delay in getting back to you. I'll add dask-ml and its xgboost extension to the image shortly. Looks like they also have an early version of a TensorFlow interface, but its pip installation is failing, so we'll have to come back to that. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#128 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADGyxXuUXeehOh6R0HMTnYOdWZbQLm6ks5s7LT6gaJpZM4Qf4C-> .