call configure_module before freeze_before_training

Author: Alan Chu

src/lightning/pytorch/callbacks/finetuning.py

@@ -274,6 +274,9 @@ def unfreeze_and_add_param_group(
 
     @override
     def setup(self, trainer: "pl.Trainer", pl_module: "pl.LightningModule", stage: str) -> None:
+        if hasattr(pl_module, "configure_model"):
+            pl_module.configure_model()
+
         self.freeze_before_training(pl_module)
 
         from lightning.pytorch.strategies import DeepSpeedStrategy