add xlm roberta classifier files (#13902)

python/sparknlp/annotator/classifier_dl/__init__.py

@@ -46,3 +46,4 @@
 from sparknlp.annotator.classifier_dl.bert_for_zero_shot_classification import *
 from sparknlp.annotator.classifier_dl.distil_bert_for_zero_shot_classification import *
 from sparknlp.annotator.classifier_dl.roberta_bert_for_zero_shot_classification import *
+from sparknlp.annotator.classifier_dl.xlm_roberta_for_zero_shot_classification import *

python/sparknlp/annotator/classifier_dl/xlm_roberta_for_zero_shot_classification.py

@@ -0,0 +1,225 @@
+#  Copyright 2017-2023 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+"""Contains classes for XlmRoBertaForZeroShotClassification."""
+
+from sparknlp.common import *
+
+
+class XlmRoBertaForZeroShotClassification(AnnotatorModel,
+                                       HasCaseSensitiveProperties,
+                                       HasBatchedAnnotate,
+                                       HasClassifierActivationProperties,
+                                       HasCandidateLabelsProperties,
+                                       HasEngine):
+    """XlmRoBertaForZeroShotClassification using a `ModelForSequenceClassification` trained on NLI (natural language
+    inference) tasks. Equivalent of `XlmRoBertaForSequenceClassification` models, but these models don't require a hardcoded
+    number of potential classes, they can be chosen at runtime. It usually means it's slower but it is much more
+    flexible.
+
+    Note that the model will loop through all provided labels. So the more labels you have, the
+    longer this process will take.
+
+    Any combination of sequences and labels can be passed and each combination will be posed as a premise/hypothesis
+    pair and passed to the pretrained model.
+
+    Pretrained models can be loaded with :meth:`.pretrained` of the companion
+    object:
+
+    >>> sequenceClassifier = XlmRoBertaForZeroShotClassification.pretrained() \\
+    ...     .setInputCols(["token", "document"]) \\
+    ...     .setOutputCol("label")
+
+    The default model is ``"xlm_roberta_large_zero_shot_classifier_xnli_anli"``, if no name is
+    provided.
+
+    For available pretrained models please see the `Models Hub
+    <https://sparknlp.orgtask=Text+Classification>`__.
+
+    To see which models are compatible and how to import them see
+    `Import Transformers into Spark NLP 🚀
+    <https://github.com/JohnSnowLabs/spark-nlp/discussions/5669>`_.
+
+    ====================== ======================
+    Input Annotation types Output Annotation type
+    ====================== ======================
+    ``DOCUMENT, TOKEN``    ``CATEGORY``
+    ====================== ======================
+
+    Parameters
+    ----------
+    batchSize
+        Batch size. Large values allows faster processing but requires more
+        memory, by default 8
+    caseSensitive
+        Whether to ignore case in tokens for embeddings matching, by default
+        True
+    configProtoBytes
+        ConfigProto from tensorflow, serialized into byte array.
+    maxSentenceLength
+        Max sentence length to process, by default 128
+    coalesceSentences
+        Instead of 1 class per sentence (if inputCols is `sentence`) output 1
+        class per document by averaging probabilities in all sentences, by
+        default False
+    activation
+        Whether to calculate logits via Softmax or Sigmoid, by default
+        `"softmax"`.
+
+    Examples
+    --------
+    >>> import sparknlp
+    >>> from sparknlp.base import *
+    >>> from sparknlp.annotator import *
+    >>> from pyspark.ml import Pipeline
+    >>> documentAssembler = DocumentAssembler() \\
+    ...     .setInputCol("text") \\
+    ...     .setOutputCol("document")
+    >>> tokenizer = Tokenizer() \\
+    ...     .setInputCols(["document"]) \\
+    ...     .setOutputCol("token")
+    >>> sequenceClassifier = XlmRoBertaForZeroShotClassification.pretrained() \\
+    ...     .setInputCols(["token", "document"]) \\
+    ...     .setOutputCol("label") \\
+    ...     .setCaseSensitive(True)
+    >>> pipeline = Pipeline().setStages([
+    ...     documentAssembler,
+    ...     tokenizer,
+    ...     sequenceClassifier
+    ... ])
+    >>> data = spark.createDataFrame([["I loved this movie when I was a child.", "It was pretty boring."]]).toDF("text")
+    >>> result = pipeline.fit(data).transform(data)
+    >>> result.select("label.result").show(truncate=False)
+    +------+
+    |result|
+    +------+
+    |[pos] |
+    |[neg] |
+    +------+
+    """
+    name = "XlmRoBertaForZeroShotClassification"
+
+    inputAnnotatorTypes = [AnnotatorType.DOCUMENT, AnnotatorType.TOKEN]
+
+    outputAnnotatorType = AnnotatorType.CATEGORY
+
+    maxSentenceLength = Param(Params._dummy(),
+                              "maxSentenceLength",
+                              "Max sentence length to process",
+                              typeConverter=TypeConverters.toInt)
+
+    configProtoBytes = Param(Params._dummy(),
+                             "configProtoBytes",
+                             "ConfigProto from tensorflow, serialized into byte array. Get with config_proto.SerializeToString()",
+                             TypeConverters.toListInt)
+
+    coalesceSentences = Param(Params._dummy(), "coalesceSentences",
+                              "Instead of 1 class per sentence (if inputCols is '''sentence''') output 1 class per document by averaging probabilities in all sentences.",
+                              TypeConverters.toBoolean)
+
+    def getClasses(self):
+        """
+        Returns labels used to train this model
+        """
+        return self._call_java("getClasses")
+
+    def setConfigProtoBytes(self, b):
+        """Sets configProto from tensorflow, serialized into byte array.
+
+        Parameters
+        ----------
+        b : List[int]
+            ConfigProto from tensorflow, serialized into byte array
+        """
+        return self._set(configProtoBytes=b)
+
+    def setMaxSentenceLength(self, value):
+        """Sets max sentence length to process, by default 128.
+
+        Parameters
+        ----------
+        value : int
+            Max sentence length to process
+        """
+        return self._set(maxSentenceLength=value)
+
+    def setCoalesceSentences(self, value):
+        """Instead of 1 class per sentence (if inputCols is '''sentence''') output 1 class per document by averaging
+        probabilities in all sentences. Due to max sequence length limit in almost all transformer models such as XlmRoBerta
+        (512 tokens), this parameter helps to feed all the sentences into the model and averaging all the probabilities
+        for the entire document instead of probabilities per sentence. (Default: true)
+
+        Parameters
+        ----------
+        value : bool
+            If the output of all sentences will be averaged to one output
+        """
+        return self._set(coalesceSentences=value)
+
+    @keyword_only
+    def __init__(self, classname="com.johnsnowlabs.nlp.annotators.classifier.dl.XlmRoBertaForZeroShotClassification",
+                 java_model=None):
+        super(XlmRoBertaForZeroShotClassification, self).__init__(
+            classname=classname,
+            java_model=java_model
+        )
+        self._setDefault(
+            batchSize=8,
+            maxSentenceLength=128,
+            caseSensitive=True,
+            coalesceSentences=False,
+            activation="softmax"
+        )
+
+    @staticmethod
+    def loadSavedModel(folder, spark_session):
+        """Loads a locally saved model.
+
+        Parameters
+        ----------
+        folder : str
+            Folder of the saved model
+            spark_session : pyspark.sql.SparkSession
+            The current SparkSession
+
+        Returns
+        -------
+        XlmRoBertaForZeroShotClassification
+            The restored model
+        """
+        from sparknlp.internal import _XlmRoBertaForZeroShotClassification
+        jModel = _XlmRoBertaForZeroShotClassification(folder, spark_session._jsparkSession)._java_obj
+        return XlmRoBertaForZeroShotClassification(java_model=jModel)
+
+    @staticmethod
+    def pretrained(name="xlm_roberta_large_zero_shot_classifier_xnli_anli", lang="xx", remote_loc=None):
+        """Downloads and loads a pretrained model.
+
+        Parameters
+        ----------
+        name : str, optional
+            Name of the pretrained model, by default
+            "xlm_roberta_large_zero_shot_classifier_xnli_anli"
+            lang : str, optional
+            Language of the pretrained model, by default "en"
+            remote_loc : str, optional
+            Optional remote address of the resource, by default None. Will use
+            Spark NLPs repositories otherwise.
+
+        Returns
+        -------
+        XlmRoBertaForZeroShotClassification
+            The restored model
+        """
+        from sparknlp.pretrained import ResourceDownloader
+        return ResourceDownloader.downloadModel(XlmRoBertaForZeroShotClassification, name, lang, remote_loc)

python/sparknlp/internal/__init__.py

@@ -536,6 +536,12 @@ def __init__(self, path, jspark):
             jspark)
 
 
+class _XlmRoBertaForZeroShotClassification(ExtendedJavaWrapper):
+    def __init__(self, path, jspark):
+        super(_XlmRoBertaForZeroShotClassification, self).__init__(
+            "com.johnsnowlabs.nlp.annotators.classifier.dl.XlmRoBertaForZeroShotClassification.loadSavedModel", path,
+            jspark)
+
 class _InstructorLoader(ExtendedJavaWrapper):
     def __init__(self, path, jspark):
         super(_InstructorLoader, self).__init__("com.johnsnowlabs.nlp.embeddings.InstructorEmbeddings.loadSavedModel", path, jspark)

python/test/annotator/classifier_dl/xlm_roberta_for_zero_shot_classification_test.py

@@ -0,0 +1,52 @@
+#  Copyright 2017-2023 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+import unittest
+import pytest
+
+from sparknlp.annotator import *
+from sparknlp.base import *
+from test.util import SparkContextForTest
+
+
+@pytest.mark.slow
+class XlmRoBertaForZeroShotClassificationTestSpec(unittest.TestCase):
+    def setUp(self):
+        self.spark = SparkContextForTest.spark
+        self.text = "I have a problem with my iphone that needs to be resolved asap!!"
+        self.inputDataset = self.spark.createDataFrame([[self.text]]) \
+            .toDF("text")
+
+    def runTest(self):
+        document_assembler = DocumentAssembler() \
+            .setInputCol("text") \
+            .setOutputCol("document")
+
+        tokenizer = Tokenizer().setInputCols("document").setOutputCol("token")
+
+        zero_shot_classifier = XlmRoBertaForZeroShotClassification \
+            .pretrained() \
+            .setInputCols(["document", "token"]) \
+            .setOutputCol("class") \
+            .setCandidateLabels(["urgent", "mobile", "travel", "movie", "music", "sport", "weather", "technology"])
+
+        pipeline = Pipeline(stages=[
+            document_assembler,
+            tokenizer,
+            zero_shot_classifier
+        ])
+
+        model = pipeline.fit(self.inputDataset)
+        model.transform(self.inputDataset).show()
+        light_pipeline = LightPipeline(model)
+        annotations_result = light_pipeline.fullAnnotate(self.text)