Use Filelock to ensure distributed barriers

see context in https://github.com/huggingface/transformers/pull/4223

src/transformers/data/datasets/language_modeling.py

@@ -4,10 +4,10 @@ import pickle
 import time
 
 import torch
+from filelock import FileLock
 from torch.utils.data.dataset import Dataset
 
 from ...tokenization_utils import PreTrainedTokenizer
-from ...trainer import torch_distributed_zero_first
 
 
 logger = logging.getLogger(__name__)
@@ -20,7 +20,7 @@ class TextDataset(Dataset):
     """
 
     def __init__(
-        self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int, overwrite_cache=False, local_rank=-1,
+        self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int, overwrite_cache=False,
     ):
         assert os.path.isfile(file_path)
 
@@ -31,9 +31,10 @@ class TextDataset(Dataset):
             directory, "cached_lm_{}_{}_{}".format(tokenizer.__class__.__name__, str(block_size), filename,),
         )
 
-        with torch_distributed_zero_first(local_rank):
-            # Make sure only the first process in distributed training processes the dataset,
-            # and the others will use the cache.
+        # Make sure only the first process in distributed training processes the dataset,
+        # and the others will use the cache.
+        lock_path = cached_features_file + ".lock"
+        with FileLock(lock_path):
 
             if os.path.exists(cached_features_file) and not overwrite_cache:
                 start = time.time()
@@ -80,7 +81,7 @@ class LineByLineTextDataset(Dataset):
     soon.
     """
 
-    def __init__(self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int, local_rank=-1):
+    def __init__(self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int):
         assert os.path.isfile(file_path)
         # Here, we do not cache the features, operating under the assumption
         # that we will soon use fast multithreaded tokenizers from the