Fix some doctests after PR 15775 (#20036)

* Add skip_special_tokens=True in some doctest

* For T5

* Fix for speech_to_text.mdx
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

docs/source/en/model_doc/speech_to_text.mdx

@@ -57,7 +57,7 @@ be installed as follows: `apt install libsndfile1-dev`
 >>> inputs = processor(ds[0]["audio"]["array"], sampling_rate=ds[0]["audio"]["sampling_rate"], return_tensors="pt")
 >>> generated_ids = model.generate(inputs["input_features"], attention_mask=inputs["attention_mask"])
 
->>> transcription = processor.batch_decode(generated_ids)
+>>> transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)
 >>> transcription
 ['mister quilter is the apostle of the middle classes and we are glad to welcome his gospel']
 ```
@@ -87,9 +87,9 @@ be installed as follows: `apt install libsndfile1-dev`
 ...     forced_bos_token_id=processor.tokenizer.lang_code_to_id["fr"],
 ... )
 
->>> translation = processor.batch_decode(generated_ids)
+>>> translation = processor.batch_decode(generated_ids, skip_special_tokens=True)
 >>> translation
-["<lang:fr> (Vidéo) Si M. Kilder est l'apossible des classes moyennes, et nous sommes heureux d'être accueillis dans son évangile."]
+["(Vidéo) Si M. Kilder est l'apossible des classes moyennes, et nous sommes heureux d'être accueillis dans son évangile."]
 ```
 
 See the [model hub](https://huggingface.co/models?filter=speech_to_text) to look for Speech2Text checkpoints.

docs/source/en/model_doc/t5.mdx

@@ -285,7 +285,7 @@ The predicted tokens will then be placed between the sentinel tokens.
 >>> sequence_ids = model.generate(input_ids)
 >>> sequences = tokenizer.batch_decode(sequence_ids)
 >>> sequences
-['<pad> <extra_id_0> park offers<extra_id_1> the<extra_id_2> park.</s>']
+['<pad><extra_id_0> park offers<extra_id_1> the<extra_id_2> park.</s>']
 ```
 
 

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -1334,7 +1334,7 @@ class Speech2TextForConditionalGeneration(Speech2TextPreTrainedModel):
 
         >>> generated_ids = model.generate(inputs=input_features)
 
-        >>> transcription = processor.batch_decode(generated_ids)[0]
+        >>> transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
         >>> transcription
         'mister quilter is the apostle of the middle classes and we are glad to welcome his gospel'
         ```"""

src/transformers/utils/doc.py

@@ -201,7 +201,7 @@ PT_QUESTION_ANSWERING_SAMPLE = r"""
     >>> answer_end_index = outputs.end_logits.argmax()
 
     >>> predict_answer_tokens = inputs.input_ids[0, answer_start_index : answer_end_index + 1]
-    >>> tokenizer.decode(predict_answer_tokens)
+    >>> tokenizer.decode(predict_answer_tokens, skip_special_tokens=True)
     {expected_output}
     ```