Warn if debug requested without TPU fixes (#6308) (#6390)

* Warn if debug requested without TPU fixes (#6308)
Check whether a PyTorch compatible TPU is available before attempting to print TPU metrics after training has completed. This way, users who apply `--debug` without reading the documentation aren't suprised by a stacktrace.

* Style

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

src/transformers/trainer.py

@@ -624,8 +624,14 @@ def train(self, model_path: Optional[str] = None):
                 train_iterator.close()
                 break
             if self.args.tpu_metrics_debug or self.args.debug:
-                # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.)
-                xm.master_print(met.metrics_report())
+                if is_torch_tpu_available():
+                    # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.)
+                    xm.master_print(met.metrics_report())
+                else:
+                    logger.warning(
+                        "You enabled PyTorch/XLA debug metrics but you don't have a TPU "
+                        "configured. Check your training configuration if this is unexpected."
+                    )
 
         if self.tb_writer:
             self.tb_writer.close()