Hokohoko

This is the base module for Hokohoko. Can be invoked either programmatically by passing an appropriately configured hokohoko.entities.Config to hokohoko.Hokohoko.run(), or by invoking in a script:

python3 -m hokohoko.Hokohoko [options]

Several options are available however to ease testing and debugging during Predictor development, so feel free to explore the options and associated documentation. However, for benchmarking, the only ones you should need to change are:

predictor       Should be set to your Predictor of choice (which
                inherits hokohoko.entities.Predictor), along with
                any required parameters.
process_count   Tune for how many cores you are willing to use.
                Hokohoko uses roughly 1GB of RAM, plus the
                Predictor's internal state, per core.

and:

assessor    The assessor class you wish to use, plus its parameters.
            Multiple assessors can be defined.

Note

The default options have been tuned from analysis of over 1000 peer-reviewed papers’ settings. If you do feel the need to use custom settings, please make sure you record them with your results, so others can compare their work with yours.

hokohoko.Hokohoko.run(config)

Runs the Hokohoko benchmark suite based on the provided config.

Parameters

config (hokohoko.entities.Config) – The global configuration options.