Fixing a bug when MlFlow try to log a torch.tensor (#29932)

* Update integration_utils.py

Add the case where a tensor with one element is log with Mlflow

* Update src/transformers/integrations/integration_utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update integration_utils.py add a whitespace

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

src/transformers/integrations/integration_utils.py

@@ -1071,6 +1071,8 @@ class MLflowCallback(TrainerCallback):
             for k, v in logs.items():
                 if isinstance(v, (int, float)):
                     metrics[k] = v
+                elif isinstance(v, torch.Tensor) and v.numel() == 1:
+                    metrics[k] = v.item()
                 else:
                     logger.warning(
                         f'Trainer is attempting to log a value of "{v}" of type {type(v)} for key "{k}" as a metric. '