Unless you're careful, GridSearch can really blow through your compute resources. That's why we're sharing some techniques to keep things lightweight. In particular, you can employ a caching mechanism such that early steps in the pipeline are cached on disk and don't get needlessly recomputed.
Video Chapters:
00:00 Introduction
01:55 Notebook example
07:54 GridSearch results
11:06 Speedup with Cache
If you're interested in joblib's cache, these docs are relevant:
joblib.readthedocs.io/en/late...
The code for all of our videos can be found on this Github repository:
github.com/probabl-ai/youtube...
The code for this specific episode can be found here:
github.com/probabl-ai/youtube...
If you're keen to see more videos like this, you can follow us over at @probabl_ai.
Негізгі бет GridSearch made faster with a Cache
Пікірлер: 4