Fix typo in speech2text2 doc (#15617)

Forward looks for inputs, not input_ids

docs/source/model_doc/speech_to_text_2.mdx

@@ -69,7 +69,7 @@ predicted token ids.
 >>> ds = ds.map(map_to_array)
 
 >>> inputs = processor(ds["speech"][0], sampling_rate=16_000, return_tensors="pt")
->>> generated_ids = model.generate(input_ids=inputs["input_values"], attention_mask=inputs["attention_mask"])
+>>> generated_ids = model.generate(inputs=inputs["input_values"], attention_mask=inputs["attention_mask"])
 
 >>> transcription = processor.batch_decode(generated_ids)
 ```