Software Open Access

Osprey v1.1.0

Robert T. McGibbon; Carlos Hernández; Matthew Harrigan; Steven Kearnes; Mohammad M. Sultan; Stanislaw Jastrzebski; Brooke Husic

Osprey is a tool for hyperparameter optimization of machine learning algorithms in Python. Hyperparameter optimization can often be an onerous process for researchers, due to time-consuming experimental replicates, non-convex objective functions, and constant tension between exploration of global parameter space and local optimization. We've designed Osprey to provide scientists with a practical, easy-to-use way of finding optimal model parameters. The software works seamlessly with scikit-learn estimators and supports many different search strategies for choosing the next set of parameters with which to evaluate a given model, including gaussian processes, tree-structured Parzen estimators, as well as random and grid search. As hyperparameter optimization is an embarrassingly parallel problem, Osprey can easily scale to hundreds of concurrent processes by executing a simple command-line program multiple times. This makes it easy to exploit large resources available in high-performance computing environments.

Osprey is actively maintained by researchers at Stanford University and other institutions around the world. While originally developed to analyze computational protein dynamics, it is applicable to any scikit-learn-compatible pipeline. The source code for Osprey is hosted on GitHub and has been archived to Zenodo. Full documentation can be found at http://msmbuilder.org/osprey.

Files (310.1 kB)
Name Size
osprey-1.1.0.zip
md5:98b754511fdefe1482209ec55fe95d3b
310.1 kB Download
44
2
views
downloads
All versions This version
Views 4410
Downloads 22
Data volume 620.3 kB620.3 kB
Unique views 4310
Unique downloads 11

Share

Cite as