Flaml/Automl - How do I pass a huge lightgbm input '.bin' file directly to lightgbm for training via flaml/automl?

[wil70]

Is there a way to pass a huge lightgbm input '.bin' file directly to lightgbm for training via flaml/automl?

The lightgbm c, cli allow me to specify directly the input file (like .bin or .csv) to use for training, it is optimize for memory. In python it requires lgb,Dataset(data='foo.bin') as far as I know

How do I do this with flaml/automl?

import lightgbm as lgb

# Set up the parameters
params = {
    'objective': 'multiclass',
    'num_classes': num_classes,
    # Add other parameters as needed
}

# Train the model with the .bin file
model = lgb.train(params, lgb.Dataset(data='path_to_file.bin'), num_boost_round=num_rounds)

Thanks for your help w

[sonichi]

This example is most close to what you need: https://github.com/microsoft/FLAML/blob/main/test/tune_example.py Please let me know if your problem can be solved by modifying that script.