o yi0@shddlmZmZmZddlmZddlmZddlm Z ddl m Z ddl m Z dZGdd d e Zd S) )AnyOptionalSequence)Literal)_bleu_score_update)_SacreBLEUTokenizer) BLEUScore)_REGEX_AVAILABLEnone13azhintlcharc seZdZUdZdZeed<dZeed<dZeed<     dd e d ed e d dede e e def fdd Zde ede e edd fddZZS)SacreBLEUScoreaCalculate `BLEU score`_ of machine translated text with one or more references. This implementation follows the behaviour of `SacreBLEU`_. The SacreBLEU implementation differs from the NLTK BLEU implementation in tokenization techniques. As input to ``forward`` and ``update`` the metric accepts the following input: - ``preds`` (:class:`~Sequence`): An iterable of machine translated corpus - ``target`` (:class:`~Sequence`): An iterable of iterables of reference corpus As output of ``forward`` and ``compute`` the metric returns the following output: - ``sacre_bleu`` (:class:`~torch.Tensor`): A tensor with the SacreBLEU Score Args: n_gram: Gram value ranged from 1 to 4 smooth: Whether to apply smoothing, see `SacreBLEU`_ tokenize: Tokenization technique to be used. Supported tokenization: ``['none', '13a', 'zh', 'intl', 'char']`` lowercase: If ``True``, BLEU score over lowercased text is calculated. kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info. weights: Weights used for unigrams, bigrams, etc. to calculate BLEU score. If not provided, uniform weights are used. Raises: ValueError: If ``tokenize`` not one of 'none', '13a', 'zh', 'intl' or 'char' ValueError: If ``tokenize`` is set to 'intl' and `regex` is not installed ValueError: If a length of a list of weights is not ``None`` and not equal to ``n_gram``. Example: >>> from torchmetrics import SacreBLEUScore >>> preds = ['the cat is on the mat'] >>> target = [['there is a cat on the mat', 'a cat is on the mat']] >>> sacre_bleu = SacreBLEUScore() >>> sacre_bleu(preds, target) tensor(0.7598) Additional References: - Automatic Evaluation of Machine Translation Quality Using Longest Common Subsequence and Skip-Bigram Statistics by Chin-Yew Lin and Franz Josef Och `Machine Translation Evolution`_ Fis_differentiableThigher_is_betterfull_state_updater Nn_gramsmoothtokenizer lowercaseweightskwargsc s\tjd|||d||tvrtdtd|d|dkr&ts&tdt|||_dS)N)rrrz*Argument `tokenize` expected to be one of z but got .rzv`'intl'` tokenization requires that `regex` is installed. Use `pip install regex` or `pip install torchmetrics[text]`.)super__init__AVAILABLE_TOKENIZERS ValueErrorr ModuleNotFoundErrorr tokenizer)selfrrrrrr __class__rP/home/ubuntu/.local/lib/python3.10/site-packages/torchmetrics/text/sacre_bleu.pyrUs  zSacreBLEUScore.__init__predstargetreturnc Cs.t|||j|j|j|j|j|j\|_|_dS)z*Update state with predictions and targets.N)r numerator denominator preds_len target_lenrr")r#r'r(rrr&updateiszSacreBLEUScore.update)rFr FN)__name__ __module__ __qualname____doc__rbool__annotations__rrintrrrfloatrrstrr. __classcell__rrr$r&r s0  0   *rN)typingrrrtyping_extensionsr!torchmetrics.functional.text.bleur'torchmetrics.functional.text.sacre_bleurtorchmetrics.text.bleurtorchmetrics.utilities.importsr rrrrrr&s