Skip to content

ml-squad/HpBandSter

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

85 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HpBandSter

a distributed Hyperband implementation on Steroids

How to install

A simple

python3 setup.py install --user

should do. It should install the dependencies automatically.

If you want to develop on the code you could install it via

python3 setup.py develop --user

How to use

Right now, there are only three example showing how to use. You can find them in the examples folder. The basic components of every run are

  1. a NameServer keeping track of the master's and workers' IPs and ports
  2. a Master (Usually BOHB or HyperBand) that coordinates the work
  3. one or more Worker instances that perform the actual computations.

The first example shows how to do computations locally on one machine with all workers launched in separate threads. The second one is closer to a usecase on a cluster (SGE in this case). It contains launching a nameserver and distributing its credentials over a shared filesystem (often found on clusters), starting a master and multiple workers. The workers don't really perform anything interesting, but hopefully it can illustrated the usage. The third example shows how live results can be logged. Use this for long and expensive runs to inspect intermediate results.

Documentation

There is no real documentation, although many methods and functions do have docstrings. There is going to be a Sphinx at some point with a bit more detail.

About

a distributed Hyperband implementation on Steroids

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%