How to use the deeppavlov.models.coreference_resolution.custom_layers.flatten function in deeppavlov

To help you get started, we’ve selected a few deeppavlov examples, based on popular ways it is used in public projects.

Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.

github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / model_on_gold.py View on Github external
Returns: word_emb, char_index, text_len, speaker_ids, genre, is_training, gold_starts, gold_ends, cluster_ids;
            it numpy tensors for placeholders (is_training - bool)
            If length of the longest sentence in the document is greater than parameter "max_training_sentences",
            the returning method calls the 'truncate_example' function.
        """
        clusters = example["clusters"]
        gold_mentions = sorted(tuple(m) for m in custom_layers.flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))
        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"]
        num_words = sum(len(s) for s in sentences)
        speakers = custom_layers.flatten(example["speakers"])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
        text_len = np.array([len(s) for s in sentences])
        doc_key = example["doc_key"]

        if self.emb_lowercase:
            for i, sentence in enumerate(sentences):
                for j, word in enumerate(sentence):
                    sentences[i][j] = word.lower()

        for i, sentence in enumerate(sentences):
            sentences[i] = list(sentences[i])
github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / new_model.py View on Github external
if isinstance(example["clusters"], tuple):
            clusters = example["clusters"][0]
        else:
            clusters = example["clusters"]

        gold_mentions = sorted(tuple(m) for m in custom_layers.flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))

        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"][0]
        num_words = sum(len(s) for s in sentences)
        speakers = custom_layers.flatten(example["speakers"][0])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
        text_len = np.array([len(s) for s in sentences])
        doc_key = example["doc_key"][0]

        if self.emb_lowercase:
            for i, sentence in enumerate(sentences):
                for j, word in enumerate(sentence):
                    sentences[i][j] = word.lower()

        for i, sentence in enumerate(sentences):
            for j, word in enumerate(sentence):
github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / new_model.py View on Github external
for tensorflow placeholders.
        Args:
            example: dict from observation
            is_training: True or False value, use as a returned parameter or flag

        Returns: word_emb, char_index, text_len, speaker_ids, genre, is_training, gold_starts, gold_ends, cluster_ids;
            it numpy tensors for placeholders (is_training - bool)
            If length of the longest sentence in the document is greater than parameter "max_training_sentences",
            the returning method calls the 'truncate_example' function.
        """
        if isinstance(example["clusters"], tuple):
            clusters = example["clusters"][0]
        else:
            clusters = example["clusters"]

        gold_mentions = sorted(tuple(m) for m in custom_layers.flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))

        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"][0]
        num_words = sum(len(s) for s in sentences)
        speakers = custom_layers.flatten(example["speakers"][0])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / model.py View on Github external
Takes a dictionary from the observation and transforms it into a set of tensors
        for tensorflow placeholders.
        Args:
            example: dict from observation
            is_training: True or False value, use as a returned parameter or flag
        Returns: word_emb, char_index, text_len, speaker_ids, genre, is_training, gold_starts, gold_ends, cluster_ids;
            it numpy tensors for placeholders (is_training - bool)
            If length of the longest sentence in the document is greater than parameter "max_training_sentences",
            the returning method calls the 'truncate_example' function.
        """
        if isinstance(example["clusters"], tuple):
            clusters = example["clusters"]
        else:
            clusters = example["clusters"]

        gold_mentions = sorted(tuple(m) for m in flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))

        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"]
        num_words = sum(len(s) for s in sentences)
        speakers = flatten(example["speakers"])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / model_on_gold.py View on Github external
def tensorize_example(self, example, is_training):
        """
        Takes a dictionary from the observation and transforms it into a set of tensors
        for tensorflow placeholders.
        Args:
            example: dict from observation
            is_training: True or False value, use as a returned parameter or flag

        Returns: word_emb, char_index, text_len, speaker_ids, genre, is_training, gold_starts, gold_ends, cluster_ids;
            it numpy tensors for placeholders (is_training - bool)
            If length of the longest sentence in the document is greater than parameter "max_training_sentences",
            the returning method calls the 'truncate_example' function.
        """
        clusters = example["clusters"]
        gold_mentions = sorted(tuple(m) for m in custom_layers.flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))
        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"]
        num_words = sum(len(s) for s in sentences)
        speakers = custom_layers.flatten(example["speakers"])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
        text_len = np.array([len(s) for s in sentences])
github deepmipt / DeepPavlov / deeppavlov / models / coreference_resolution / model.py View on Github external
if isinstance(example["clusters"], tuple):
            clusters = example["clusters"]
        else:
            clusters = example["clusters"]

        gold_mentions = sorted(tuple(m) for m in flatten(clusters))
        gold_mention_map = {m: i for i, m in enumerate(gold_mentions)}
        cluster_ids = np.zeros(len(gold_mentions))

        for cluster_id, cluster in enumerate(clusters):
            for mention in cluster:
                cluster_ids[gold_mention_map[tuple(mention)]] = cluster_id

        sentences = example["sentences"]
        num_words = sum(len(s) for s in sentences)
        speakers = flatten(example["speakers"])

        assert num_words == len(speakers)

        max_sentence_length = max(len(s) for s in sentences)
        max_word_length = max(max(max(len(w) for w in s) for s in sentences), max(self.filter_widths))
        char_index = np.zeros([len(sentences), max_sentence_length, max_word_length])
        text_len = np.array([len(s) for s in sentences])
        doc_key = example["doc_key"]

        if self.emb_lowercase:
            for i, sentence in enumerate(sentences):
                for j, word in enumerate(sentence):
                    sentences[i][j] = word.lower()

        for i, sentence in enumerate(sentences):
            for j, word in enumerate(sentence):