fix: use semantic version comparison for PyTorch scheduler compatibility (#2094)

Author: SunsetWolf

qlib/contrib/model/pytorch_nn.py

@@ -10,6 +10,7 @@
 import gc
 import numpy as np
 import pandas as pd
+from packaging import version
 from typing import Callable, Optional, Text, Union
 from sklearn.metrics import roc_auc_score, mean_squared_error
 
@@ -148,7 +149,7 @@ def __init__(
         if scheduler == "default":
             # In torch version 2.7.0, the verbose parameter has been removed. Reference Link:
             # https://github.com/pytorch/pytorch/pull/147301/files#diff-036a7470d5307f13c9a6a51c3a65dd014f00ca02f476c545488cd856bea9bcf2L1313
-            if str(torch.__version__).split("+", maxsplit=1)[0] <= "2.6.0":
+            if version.parse(str(torch.__version__).split("+", maxsplit=1)[0]) <= version.parse("2.6.0"):
                 # Reduce learning rate when loss has stopped decrease
                 self.scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(  # pylint: disable=E1123
                     self.train_optimizer,