Safe import of LRScheduler (#29919)

* Safe import of LRScheduler

* Update src/transformers/trainer_pt_utils.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/trainer_pt_utils.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Fix up

---------

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

src/transformers/trainer_pt_utils.py

@@ -34,13 +34,18 @@
 import torch
 import torch.distributed as dist
 from torch import nn
-from torch.optim.lr_scheduler import LRScheduler
 from torch.utils.data import Dataset, IterableDataset, RandomSampler, Sampler
 from torch.utils.data.distributed import DistributedSampler
 
 from .integrations.deepspeed import is_deepspeed_zero3_enabled
 from .tokenization_utils_base import BatchEncoding
-from .utils import is_sagemaker_mp_enabled, is_torch_xla_available, is_training_run_on_sagemaker, logging
+from .utils import (
+    is_sagemaker_mp_enabled,
+    is_torch_available,
+    is_torch_xla_available,
+    is_training_run_on_sagemaker,
+    logging,
+)
 
 
 if is_training_run_on_sagemaker():
@@ -49,6 +54,15 @@
 if is_torch_xla_available():
     import torch_xla.core.xla_model as xm
 
+if is_torch_available():
+    from .pytorch_utils import is_torch_greater_or_equal_than_2_0
+
+    if is_torch_greater_or_equal_than_2_0:
+        from torch.optim.lr_scheduler import LRScheduler
+    else:
+        from torch.optim.lr_scheduler import _LRScheduler as LRScheduler
+
+
 # this is used to suppress an undesired warning emitted by pytorch versions 1.4.2-1.7.0
 try:
     from torch.optim.lr_scheduler import SAVE_STATE_WARNING