[breaking] Change DMatrix construction to be distributed #9623
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
trivialfis
reviewed
Oct 3, 2023
include/xgboost/data.h
Outdated
| @@ -559,8 +559,9 @@ class DMatrix { | |||
| * | |||
| * \param uri The URI of input. | |||
| * \param silent Whether print information during loading. | |||
| * \param data_split_mode In distributed mode, split the input according this mode; otherwise, | |||
| * it's just an indicator on how the input was split beforehand. | |||
| * \param data_split_mode In distributed mode, if the data split mode is row, split the input by | |||
There was a problem hiding this comment.
Let's remove the row splitting as well. I can help with that if it doesn't conflict with your PR.
There was a problem hiding this comment.
Are you sure that's what we want to do? In the text readers there is logic to optimize for reading a portion of the file, so in terms of performance it's probably not that far off from reading a whole file. This behavior has been around since the beginning of xgboost, I wonder how many people actually rely on it.
There was a problem hiding this comment.
Thank you for removing the old code, could you please help fix the lint errors?
There was a problem hiding this comment.
I will fix it in a separated PR for backporting: #9634 .
rongou
changed the title
trivialfis
changed the title
trivialfis
approved these changes
Oct 10, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently when we load a DMatrix from text files, we assume each worker has access to the whole dataset. Data is split by row or column depending on the
data_split_modeparameter. This is different from vertical federated learning, where each worker only has access to its own subset of data.With this change we now assume each worker only loads its own subset of data. For column-wise split, after data loading the columns are re-indexed to reflect the global view.
In support of #9619