Ray’s new release provides an optimised Ray-on-Ray library for distributed data parallel training, which allows developers to train models on multiple machines without rewriting their code.
The new release includes a new AutoScaling service that can automatically adjust the number of workers used to train a model in response to changes in resource usage. AutoScaling also provides a history of training runs and performance data for debugging purposes.
Ray 2.0 is available now on GitHub.
Scikit-learn 0.23.1 released with new features and bug fixes
Scikit-learn, the Python machine learning library, has reached version 0.23.1 with improvements and bug fixes. New features include a new clustering algorithm, Spectral Clustering. The algorithm works by projecting data into a lower dimensional space where it can be clustered more easily.
Other new features include support for missing values in regression and classification models, as well as a new RandomizedLassoCV class that can be used to select features. A full list of changes and bug fixes can be found in the release notes.
Scikit-learn 0.23.1 is available now on GitHub.
More posts
