Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[jvm-packages] remove default parameters #7938

Merged
merged 2 commits into from
May 28, 2022

Conversation

wbo4958
Copy link
Contributor

@wbo4958 wbo4958 commented May 25, 2022

This PR is to remove the most booster default parameters, which will be defaulted by xgboost native.

@wbo4958
Copy link
Contributor Author

wbo4958 commented May 26, 2022

@trivialfis could you help to check the python test CI?

@trivialfis
Copy link
Member

Please ignore it for now. The hypothesis package is updated and caught the error where xgboost failed to parse some parameters in double precision.

Copy link
Member

@trivialfis trivialfis left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you please set this parameter to true

bool validate_parameters {false};
and check if there's any warning during training? If there is no warning then you can pass it from the scala package to enable the validation.

@wbo4958
Copy link
Contributor Author

wbo4958 commented May 26, 2022

Could you please set this parameter to true

bool validate_parameters {false};

and check if there's any warning during training? If there is no warning then you can pass it from the scala package to enable the validation.

Yeah, there are some WARNINGs,

Parameters: { "allow_non_zero_for_missing", "dmlc_worker_connect_retry", "features_col", "handle_invalid", "label_col", "missing", "num_early_stopping_rounds", "num_round", "num_workers", "prediction_col", "probability_col", "rabit_ring_reduce_threshold", "rabit_timeout", "raw_prediction_col", "silent", "tracker_conf", "train_test_ratio", "tree_limit", "use_external_memory" } might not be used.

Most parameters unused are coming from the JVM parameter default values, and we should delete these parameters before passing into native. I also have filed an item (AKA an followup) to unify these parameters.

@trivialfis
Copy link
Member

Hmm, rabit timeout is useful.

@trivialfis
Copy link
Member

Hmm, rabit timeout is useful.

I take it back. It's not useful for XGBoost training, this should be removed after rabit initialization.

@trivialfis
Copy link
Member

@wbo4958 Could you please rebase the branch to master?

@trivialfis trivialfis merged commit fbc3d86 into dmlc:master May 28, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants