Check max_bin, etc. match config when using binary

[cyfdecyf]

To avoid binary file with non-matching parameters from being used.

[ghost]

All CLA requirements met.

[guolinke]

Thank you @cyfdecyf

[StrikerRUS]

@guolinke What about other Dataset params from https://lightgbm.readthedocs.io/en/latest/Parameters.html#dataset-parameters, e.g. max_bin_by_feature, bin_construct_sample_cnt? Should they also be checked?

[guolinke]

@StrikerRUS yes, thank you.
@cyfdecyf can you add these parameters too?

[cyfdecyf]

@guolinke OK. I will add checking for those parameters too.

[StrikerRUS]

I will add checking for those parameters too.

I think there are more params in Dataset Parameters section that can be checked.

Also, please refer to

LightGBM/src/c_api.cpp

Line 186 in 3e8e24b

static void CheckDatasetResetConfig(

.

[cyfdecyf]

I focused only on checking parameters that are stored in Dataset

LightGBM/include/LightGBM/dataset.h

Line 636 in 3e8e24b

std::string data_filename_;

What's the desired way to check parameters that's not in Dataset now? e.g. feature_pre_filter, data_random_seed etc.
Storing them in the binary file and only load them for checking? That would be a bigger change than this PR.

I will add checking for those parameters too.

I think there are more params in Dataset Parameters section that can be checked.

[guolinke]

@cyfdecyf yes, I think only check the stored parameter in the Dataset binary object is enough.
you can ping me once the PR is ready.

[cyfdecyf]

@guolinke I've already added checking for max_bin_by_feature, bin_construct_sample_cnt in 9d19083.

I've checked parameters listed in https://lightgbm.readthedocs.io/en/latest/Parameters.html#dataset-parameters against private members in Dataset, but not sure if there's any other parameters that still need to be checked.

[guolinke]

current implementation looks good to me. ping @StrikerRUS to confirm.

[StrikerRUS]

OK, I agree!

Thanks a lot @cyfdecyf for contribution!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.