Russian BertForSequenceClassification Cased model (from IlyaGusev)

Description

Pretrained BertForSequenceClassification model, adapted from Hugging Face and curated to provide scalability and production-readiness using Spark NLP. rubertconv_toxic_clf is a Russian model originally trained by IlyaGusev.

Predicted Entities

toxic, neutral

Live Demo Open in Colab Download Copy S3 URI

How to use

documentAssembler = DocumentAssembler() \
    .setInputCol("text") \
    .setOutputCol("document")

tokenizer = Tokenizer() \
    .setInputCols("document") \
    .setOutputCol("token")

seq_classifier = BertForSequenceClassification.pretrained("bert_classifier_rubertconv_toxic_clf","ru") \
    .setInputCols(["document", "token"]) \
    .setOutputCol("class")

pipeline = Pipeline(stages=[documentAssembler, tokenizer, seq_classifier])

data = spark.createDataFrame([["PUT YOUR STRING HERE"]]).toDF("text")

result = pipeline.fit(data).transform(data)

val documentAssembler = new DocumentAssembler()
      .setInputCols(Array("text"))
      .setOutputCols(Array("document"))

val tokenizer = new Tokenizer()
    .setInputCols("document")
    .setOutputCol("token")

val seq_classifier = BertForSequenceClassification.pretrained("bert_classifier_rubertconv_toxic_clf","ru")
    .setInputCols(Array("document", "token"))
    .setOutputCol("class")

val pipeline = new Pipeline().setStages(Array(documentAssembler, tokenizer, seq_classifier))

val data = Seq("PUT YOUR STRING HERE").toDS.toDF("text")

val result = pipeline.fit(data).transform(data)

import nlu
nlu.load("ru.classify.bert").predict("""PUT YOUR STRING HERE""")

Model Information

References

• https://huggingface.co/IlyaGusev/rubertconv_toxic_clf
• https://www.kaggle.com/blackmoon/russian-language-toxic-comments
• https://www.kaggle.com/alexandersemiletov/toxic-russian-comments
• https://toloka.ai/ru/datasets
• https://github.com/Koziev/NLP_Datasets/blob/master/Conversations/Data