workaround documentation rendering bug (#21189)

src/transformers/feature_extraction_sequence_utils.py

@@ -107,7 +107,7 @@ class SequenceFeatureExtractor(FeatureExtractionMixin):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific feature_extractor's default.
@@ -250,7 +250,7 @@ class SequenceFeatureExtractor(FeatureExtractionMixin):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """
@@ -309,7 +309,7 @@ class SequenceFeatureExtractor(FeatureExtractionMixin):
             max_length: maximum length of the returned list and optionally padding length (see below)
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             truncation:
                 (optional) Activates truncation to cut input sequences longer than `max_length` to `max_length`.
         """

src/transformers/models/layoutlmv2/tokenization_layoutlmv2.py

@@ -100,7 +100,7 @@ LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -1288,7 +1288,7 @@ class LayoutLMv2Tokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/layoutlmv2/tokenization_layoutlmv2_fast.py

@@ -710,7 +710,7 @@ class LayoutLMv2TokenizerFast(PreTrainedTokenizerFast):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/layoutlmv3/tokenization_layoutlmv3.py

@@ -97,7 +97,7 @@ LAYOUTLMV3_ENCODE_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -146,7 +146,7 @@ LAYOUTLMV3_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -1420,7 +1420,7 @@ class LayoutLMv3Tokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/layoutlmv3/tokenization_layoutlmv3_fast.py

@@ -762,7 +762,7 @@ class LayoutLMv3TokenizerFast(PreTrainedTokenizerFast):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/layoutxlm/tokenization_layoutxlm.py

@@ -82,7 +82,7 @@ LAYOUTXLM_ENCODE_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -1118,7 +1118,7 @@ class LayoutXLMTokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/layoutxlm/tokenization_layoutxlm_fast.py

@@ -85,7 +85,7 @@ LAYOUTXLM_ENCODE_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -674,7 +674,7 @@ class LayoutXLMTokenizerFast(PreTrainedTokenizerFast):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/luke/tokenization_luke.py

@@ -1440,7 +1440,7 @@ class LukeTokenizer(PreTrainedTokenizer):
                 The maximum length of the entity sequence.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific tokenizer's default, defined by the `return_outputs` attribute. [What are attention
@@ -1584,7 +1584,7 @@ class LukeTokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/markuplm/tokenization_markuplm.py

@@ -96,7 +96,7 @@ MARKUPLM_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -1392,7 +1392,7 @@ class MarkupLMTokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/markuplm/tokenization_markuplm_fast.py

@@ -806,7 +806,7 @@ class MarkupLMTokenizerFast(PreTrainedTokenizerFast):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/mctct/feature_extraction_mctct.py

@@ -275,7 +275,7 @@ class MCTCTFeatureExtractor(SequenceFeatureExtractor):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific feature_extractor's default.

src/transformers/models/mluke/tokenization_mluke.py

@@ -1238,7 +1238,7 @@ class MLukeTokenizer(PreTrainedTokenizer):
                 The maximum length of the entity sequence.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific tokenizer's default, defined by the `return_outputs` attribute. [What are attention
@@ -1383,7 +1383,7 @@ class MLukeTokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/speech_to_text/feature_extraction_speech_to_text.py

@@ -160,7 +160,7 @@ class Speech2TextFeatureExtractor(SequenceFeatureExtractor):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific feature_extractor's default.

src/transformers/models/tapas/tokenization_tapas.py

@@ -223,7 +223,7 @@ TAPAS_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING = r"""
                 which it will tokenize. This is useful for NER or token classification.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -1852,7 +1852,7 @@ class TapasTokenizer(PreTrainedTokenizer):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/models/tapex/tokenization_tapex.py

@@ -106,7 +106,7 @@ TAPEX_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING = r"""
                 argument defines the number of overlapping tokens.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 

src/transformers/models/wav2vec2/feature_extraction_wav2vec2.py

@@ -136,7 +136,7 @@ class Wav2Vec2FeatureExtractor(SequenceFeatureExtractor):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific feature_extractor's default.

src/transformers/models/wav2vec2/tokenization_wav2vec2.py

@@ -88,7 +88,7 @@ WAV2VEC2_KWARGS_DOCSTRING = r"""
                 length (like XLNet) truncation/padding to a maximum length will be deactivated.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 

src/transformers/models/whisper/feature_extraction_whisper.py

@@ -240,7 +240,7 @@ class WhisperFeatureExtractor(SequenceFeatureExtractor):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
+                `>= 7.5` (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific feature_extractor's default.

src/transformers/tokenization_utils_base.py

@@ -1343,7 +1343,7 @@ ENCODE_KWARGS_DOCSTRING = r"""
                 which it will tokenize. This is useful for NER or token classification.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
             return_tensors (`str` or [`~utils.TensorType`], *optional*):
                 If set, will return tensors instead of list of python integers. Acceptable values are:
 
@@ -2902,7 +2902,7 @@ class PreTrainedTokenizerBase(SpecialTokensMixin, PushToHubMixin):
                 If set will pad the sequence to a multiple of the provided value.
 
                 This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask (`bool`, *optional*):
                 Whether to return the attention mask. If left to the default, will return the attention mask according
                 to the specific tokenizer's default, defined by the `return_outputs` attribute.
@@ -3339,7 +3339,7 @@ class PreTrainedTokenizerBase(SpecialTokensMixin, PushToHubMixin):
                     - 'right': pads on the right of the sequences
             pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                 This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
-                >= 7.5 (Volta).
+                `>= 7.5` (Volta).
             return_attention_mask:
                 (optional) Set to False to avoid returning attention mask (default: set to model specifics)
         """

src/transformers/tokenization_utils_fast.py

@@ -346,7 +346,7 @@ class PreTrainedTokenizerFast(PreTrainedTokenizerBase):
                 The stride to use when handling overflow.
             pad_to_multiple_of (`int`, *optional*):
                 If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
-                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
+                the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta).
         """
         _truncation = self._tokenizer.truncation
         _padding = self._tokenizer.padding