dle additional num_candidate axis. (batch_size, num_candidates, text)
            2. Always pad the sequences to *max_length*.
            3. Must specify *max_length* in order to stack packs of candidates into a batch.

            - single sequence: `[CLS] X [SEP]`
            - pair of sequences: `[CLS] A [SEP] B [SEP]`

        Args:
            text (`List[List[str]]`):
                The batch of sequences to be encoded. Each sequence must be in this format: (batch_size,
                num_candidates, text).
            text_pair (`List[List[str]]`, *optional*):
                The batch of sequences to be encoded. Each sequence must be in this format: (batch_size,
                num_candidates, text).
            **kwargs:
                Keyword arguments of the __call__ method.

        Returns:
            [`BatchEncoding`]: Encoded text or text pair.

        Example:

        ```python
        >>> from transformers import RealmTokenizer

        >>> # batch_size = 2, num_candidates = 2
        >>> text = [["Hello world!", "Nice to meet you!"], ["The cute cat.", "The adorable dog."]]

        >>> tokenizer = RealmTokenizer.from_pretrained("google/realm-cc-news-pretrained-encoder")
        >>> tokenized_text = tokenizer.batch_encode_candidates(text, max_length=10, return_tensors="pt")
        ```Ú