Bug: `training_args.py` fix missing import with accelerate with version...

Bug: `training_args.py` fix missing import with accelerate with version `accelerate==0.20.1` (#28171)

* fix-accelerate-version

* updated with exported ACCELERATE_MIN_VERSION,

* update string in ACCELERATE_MIN_VERSION

src/transformers/training_args.py

@@ -36,6 +36,7 @@ from .trainer_utils import (
     SchedulerType,
 )
 from .utils import (
+    ACCELERATE_MIN_VERSION,
     ExplicitEnum,
     cached_property,
     is_accelerate_available,
@@ -1837,9 +1838,10 @@ class TrainingArguments:
         requires_backends(self, ["torch"])
         logger.info("PyTorch: setting up devices")
         if not is_sagemaker_mp_enabled():
-            if not is_accelerate_available(min_version="0.20.1"):
+            if not is_accelerate_available():
                 raise ImportError(
-                    "Using the `Trainer` with `PyTorch` requires `accelerate>=0.20.1`: Please run `pip install transformers[torch]` or `pip install accelerate -U`"
+                    f"Using the `Trainer` with `PyTorch` requires `accelerate>={ACCELERATE_MIN_VERSION}`: "
+                    "Please run `pip install transformers[torch]` or `pip install accelerate -U`"
                 )
             AcceleratorState._reset_state(reset_partial_state=True)
         self.distributed_state = None

src/transformers/utils/__init__.py

@@ -90,6 +90,7 @@ from .hub import (
     try_to_load_from_cache,
 )
 from .import_utils import (
+    ACCELERATE_MIN_VERSION,
     ENV_VARS_TRUE_AND_AUTO_VALUES,
     ENV_VARS_TRUE_VALUES,
     TORCH_FX_REQUIRED_VERSION,