fixed an issue with eval forcing to have a val_dataloader in config, …

…while for train it wasnt required (#1823)

Co-authored-by: Eugene Khvedchenya <ekhvedchenya@gmail.com>

src/super_gradients/training/sg_trainer/sg_trainer.py

@@ -353,7 +353,9 @@ def evaluate_from_recipe(cls, cfg: DictConfig) -> Tuple[nn.Module, Tuple]:
 
         # INSTANTIATE DATA LOADERS
         val_dataloader = dataloaders.get(
-            name=cfg.val_dataloader, dataset_params=cfg.dataset_params.val_dataset_params, dataloader_params=cfg.dataset_params.val_dataloader_params
+            name=get_param(cfg, "val_dataloader"),
+            dataset_params=cfg.dataset_params.val_dataset_params,
+            dataloader_params=cfg.dataset_params.val_dataloader_params,
         )
 
         if cfg.checkpoint_params.checkpoint_path is None: