Replace BloomTokenizer by BloomTokenizerFast in doc (#18005)

src/transformers/models/bloom/modeling_bloom.py

@@ -38,7 +38,7 @@ logger = logging.get_logger(__name__)
 
 _CHECKPOINT_FOR_DOC = "bigscience/Bloom"
 _CONFIG_FOR_DOC = "BloomConfig"
-_TOKENIZER_FOR_DOC = "BloomTokenizer"
+_TOKENIZER_FOR_DOC = "BloomTokenizerFast"
 
 BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST = [
     "bigscience/bigscience-small-testing",
@@ -626,7 +626,7 @@ BLOOM_INPUTS_DOCSTRING = r"""
             If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
             `input_ids`.
 
-            Indices can be obtained using [`BloomTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            Indices can be obtained using [`BloomTokenizerFast`]. See [`PreTrainedTokenizer.encode`] and
             [`PreTrainedTokenizer.__call__`] for details.
 
             [What are input IDs?](../glossary#input-ids)