ReduceLROnPlateau unexpected keyword argument

[Samoed]

🐛 Bug

Environment

1. lightautoml version: 0.4.1
2. python version: 3.11
3. OS: linux
4. pip freeze file:

To Reproduce

Steps to reproduce the behavior:

from lightautoml.automl.presets.text_presets import TabularNLPAutoML
from lightautoml.tasks import Task

automl = TabularNLPAutoML(task = Task(name = 'multiclass', metric = 'f1_macro'))
oof_preds = automl.fit_predict(train_df, roles = {'target': 'label'}).data

ReduceLROnPlateau.__init__() got an unexpected keyword argument 'verbose'
INFO3:lightautoml.ml_algo.utils:Traceback (most recent call last):
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/ml_algo/utils.py", line 77, in tune_and_fit_predict
    preds = ml_algo.fit_predict(train_valid)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/ml_algo/dl_model.py", line 556, in fit_predict
    return super().fit_predict(train_valid_iterator)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/ml_algo/base.py", line 275, in fit_predict
    model, pred = self.fit_predict_single_fold(train, valid)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/ml_algo/dl_model.py", line 582, in fit_predict_single_fold
    val_pred = model.fit(dataloaders)
               ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/text/trainer.py", line 430, in fit
    self._init()
  File "/home/samoed/Desktop/AutoIntent/.venv/lib/python3.11/site-packages/lightautoml/text/trainer.py", line 346, in _init
    self.scheduler = self.sch(self.optimizer, **self.scheduler_params) if self.sch is not None else None
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: ReduceLROnPlateau.__init__() got an unexpected keyword argument 'verbose'

[Uptimolli]

Hi, thank you for using our library.

The reason for this error is that in the new version of torch (torch==2.7.0) the verbose parameter is removed, we will add a bugfix in the next release, but in the meantime you can use an earlier version of torch (e.g. torch==2.6.0) with latest LightAutoML version.

Sincerely, LightAutoML team.