dmlc / dmlc/xgboost

[Python] How to reuse the .gradient_index.page cache files for ExtMemQuantileDMatrix?

Open
#11,415 2 comments 0 reactions 0 assignees View on GitHub
feature-request
Dominant language
C++
Stars
28.8k
Forks
8.9k
Avg merge
1d 12h
Merged PRs (30d)
54

Description

I have a custom data iterator that I pass into the constructor for `ExtMemQuantileDMatrix`, which when called produces huge cache files with names of the form `cache-*.gradient_index.page`. Is there a way to reuse these cache files between runs, so that the same `ExtMemQuantileDMatrix` does not need to be recreated from scratch?

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with the Python ExtMemQuantileDMatrix constructor and trace how the custom data iterator creates cache-*.gradient_index.page files. Determine whether those files can safely be reused between runs; done should be a documented reuse path or a clearly documented limitation, with coverage for repeated construction if the project has an applicable test location.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.