Github user jkbradley commented on the issue:
https://github.com/apache/spark/pull/18313
Oh, you're right; I overlooked that it only holds all of the models for a
single split. In that case, I agree it could be problematic to keep all in
memory by default. How does this sound then: We can do 2 separate PRs, each of
which adds a separate option:
1. Add an option for keeping the models in memory so that they are
available as a data field in the Model object after training. This caters to
smaller use cases, focusing on ease-of-use.
2. (your PR) Add an option to dump models to a directory, not keeping them
in memory. This caters to big use cases, focusing on scalability.
What do you think? If this sounds good, feel free to reopen your current
PR.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
