How to configure classification labels (based on return ranking) and adapt model training for classification tasks

[scnu001]

❓ Questions and Help

How can I configure the label in the task config to generate discrete classification labels instead of continuous regression values? Specifically, I want to discretize the returns based on their cross-sectional ranking.
Once the labels are discrete, how should I configure the model and loss function to perform classification training?
And How should the backtesting module be adapted for classification outputs?

[scnu001]

I find we can add binary in model loss like this '''model_config = {
"class": "LGBModel",
"module_path": "qlib.contrib.model.gbdt",
"kwargs": {
"loss": "binary",
"colsample_bytree": 0.8879,
"learning_rate": 0.0421,
"subsample": 0.8789,
"lambda_l1": 205.6999,
"lambda_l2": 580.9768,
"max_depth": 8,
"num_leaves": 210,
"num_threads": 20,
"seed": 688
}
}''' but when i check the dataset like this code handler = dataset.handler
print(handler.get_cols())
print(handler.get_cols('feature'))
print(handler.get_cols('label'))
print(df_train['LABEL0'].head()) i find that there are only label0 yield so how can model classify train. Does model use yield if >0 or not to class?