Skip to content
Snippets Groups Projects
Commit 0224731f authored by Yifan Zhao's avatar Yifan Zhao
Browse files

Use 'kept' configs consistently throughout

parent 9261e5ec
No related branches found
No related tags found
No related merge requests found
...@@ -125,15 +125,17 @@ We will be using the term QoS throughout the tutorials. ...@@ -125,15 +125,17 @@ We will be using the term QoS throughout the tutorials.
:py:meth:`tuner.tune <predtuner.modeledapp.ApproxModeledTuner.tune>` :py:meth:`tuner.tune <predtuner.modeledapp.ApproxModeledTuner.tune>`
is the main method for running a tuning session. is the main method for running a tuning session.
It accepts a few parameters which controls the behavior of tuning. It accepts a few parameters which controls the behavior of tuning.
`max_iter` defines the number of iterations to use in autotuning.
Within 1000 iterations, PredTuner should find about 200 valid configurations. * `qos_keep_threshold` decides the QoS threshold above which the found configuration is kept.
PredTuner will also automatically mark out `Pareto-optimal These are called the "kept" configurations and are accessible from `tuner.kept_configs`.
<https://en.wikipedia.org/wiki/Pareto_efficiency>`_
configurations. * `max_iter` defines the number of iterations to use in autotuning.
These are called "best" configurations (`tuner.best_configs`), Within 1000 iterations, PredTuner should be about to find about 200 "kept" configurations.
in contrast to "valid" configurations which are the configurations that satisfy our accuracy requirements
(`tuner.kept_configs`). * PredTuner will also automatically mark out
`take_best_n` allows taking some extra close-optimal configurations in addition to Pareto-optimal ones. `Pareto-optimal <https://en.wikipedia.org/wiki/Pareto_efficiency>`_ configurations.
These are called "best" configurations (`tuner.best_configs`)
`take_best_n` allows taking some extra close-optimal configurations in addition to Pareto-optimal ones.
1000 iterations is for demonstration; in practice, 1000 iterations is for demonstration; in practice,
at least 10000 iterations are necessary on VGG16-sized models to converge to a set of good configurations. at least 10000 iterations are necessary on VGG16-sized models to converge to a set of good configurations.
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment