Fixes for the documentation (#13361)

src/transformers/configuration_utils.py

@@ -87,7 +87,7 @@ class PretrainedConfig(PushToHubMixin):
             Whether cross-attention layers should be added to the model. Note, this option is only relevant for models
             that can be used as decoder models within the `:class:~transformers.EncoderDecoderModel` class, which
             consists of all models in ``AUTO_MODELS_FOR_CAUSAL_LM``.
-        tie_encoder_decoder (:obj:`bool`, `optional`, defaults to :obj:`False`)
+        tie_encoder_decoder (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Whether all encoder weights should be tied to their equivalent decoder weights. This requires the encoder
             and decoder model to have the exact same parameter names.
         prune_heads (:obj:`Dict[int, List[int]]`, `optional`, defaults to :obj:`{}`):

src/transformers/data/__init__.py

@@ -16,6 +16,16 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from .data_collator import (
+    DataCollatorForLanguageModeling,
+    DataCollatorForPermutationLanguageModeling,
+    DataCollatorForSeq2Seq,
+    DataCollatorForSOP,
+    DataCollatorForTokenClassification,
+    DataCollatorForWholeWordMask,
+    DataCollatorWithPadding,
+    default_data_collator,
+)
 from .metrics import glue_compute_metrics, xnli_compute_metrics
 from .processors import (
     DataProcessor,

src/transformers/deepspeed.py

@@ -50,7 +50,7 @@ class HfDeepSpeedConfig:
     values: ``"auto"``. Without this special logic the DeepSpeed configuration is not modified in any way.
 
     Args:
-        config_file_or_dict (:obj:`Union[str, Dict]`) - path to DeepSpeed config file or dict.
+        config_file_or_dict (:obj:`Union[str, Dict]`): path to DeepSpeed config file or dict.
 
     """
 

src/transformers/generation_tf_utils.py

@@ -1535,10 +1535,14 @@ def tf_top_k_top_p_filtering(logits, top_k=0, top_p=1.0, filter_value=-float("In
 
     Args:
         logits: logits distribution shape (batch size, vocabulary size)
-        if top_k > 0: keep only top k tokens with highest probability (top-k filtering).
-        if top_p < 1.0: keep the top tokens with cumulative probability >= top_p (nucleus filtering).
-            Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
-        Make sure we keep at least min_tokens_to_keep per batch example in the output
+        top_k (:obj:`int`, `optional`, defaults to 0):
+            If > 0, only keep the top k tokens with highest probability (top-k filtering)
+        top_p (:obj:`float`, `optional`, defaults to 1.0):
+            If < 1.0, only keep the top tokens with cumulative probability >= top_p (nucleus filtering). Nucleus
+            filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
+        min_tokens_to_keep (:obj:`int`, `optional`, defaults to 1):
+            Minimumber of tokens we keep per batch example in the output.
+
     From: https://gist.github.com/thomwolf/1a5a29f6962089e871b94cbd09daf317
     """
     logits_shape = shape_list(logits)

src/transformers/generation_utils.py

@@ -2557,10 +2557,14 @@ def top_k_top_p_filtering(
 
     Args:
         logits: logits distribution shape (batch size, vocabulary size)
-        if top_k > 0: keep only top k tokens with highest probability (top-k filtering).
-        if top_p < 1.0: keep the top tokens with cumulative probability >= top_p (nucleus filtering).
-            Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
-        Make sure we keep at least min_tokens_to_keep per batch example in the output
+        top_k (:obj:`int`, `optional`, defaults to 0):
+            If > 0, only keep the top k tokens with highest probability (top-k filtering)
+        top_p (:obj:`float`, `optional`, defaults to 1.0):
+            If < 1.0, only keep the top tokens with cumulative probability >= top_p (nucleus filtering). Nucleus
+            filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
+        min_tokens_to_keep (:obj:`int`, `optional`, defaults to 1):
+            Minimumber of tokens we keep per batch example in the output.
+
     From: https://gist.github.com/thomwolf/1a5a29f6962089e871b94cbd09daf317
     """
     if top_k > 0:

src/transformers/models/speech_to_text/configuration_speech_to_text.py

@@ -78,7 +78,7 @@ class Speech2TextConfig(PretrainedConfig):
             Whether or not the model should return the last key/values attentions (not used by all models).
         max_source_positions (:obj:`int`, `optional`, defaults to 6000):
             The maximum sequence length of log-mel filter-bank features that this model might ever be used with.
-        max_target_positions: (:obj:`int`, `optional`, defaults to 1024):
+        max_target_positions (:obj:`int`, `optional`, defaults to 1024):
             The maximum sequence length that this model might ever be used with. Typically set this to something large
             just in case (e.g., 512 or 1024 or 2048).
         num_conv_layers (:obj:`int`, `optional`, defaults to 2):
@@ -95,7 +95,7 @@ class Speech2TextConfig(PretrainedConfig):
         input_channels (:obj:`int`, `optional`, defaults to 1):
             An integer specifying number of input channels of the input feature vector.
 
-        Example::
+    Example::
 
         >>> from transformers import Speech2TextModel, Speech2TextConfig
 

src/transformers/pipelines/__init__.py

@@ -306,10 +306,10 @@ def pipeline(
 
             - :obj:`"feature-extraction"`: will return a :class:`~transformers.FeatureExtractionPipeline`.
             - :obj:`"text-classification"`: will return a :class:`~transformers.TextClassificationPipeline`.
-            - :obj:`"sentiment-analysis"`: (alias of :obj:`"text-classification") will return a
+            - :obj:`"sentiment-analysis"`: (alias of :obj:`"text-classification"`) will return a
               :class:`~transformers.TextClassificationPipeline`.
             - :obj:`"token-classification"`: will return a :class:`~transformers.TokenClassificationPipeline`.
-            - :obj:`"ner"` (alias of :obj:`"token-classification"): will return a
+            - :obj:`"ner"` (alias of :obj:`"token-classification"`): will return a
               :class:`~transformers.TokenClassificationPipeline`.
             - :obj:`"question-answering"`: will return a :class:`~transformers.QuestionAnsweringPipeline`.
             - :obj:`"fill-mask"`: will return a :class:`~transformers.FillMaskPipeline`.