DistilBERT Sequence Classification - Urdu IMDB (distilbert_base_sequence_classifier_imdb)

Description

DistilBERT Model with sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for multi-class document classification tasks.

distilbert_base_sequence_classifier_imdb is a fine-tuned DistilBERT model that is ready to be used for Sequence Classification tasks such as sentiment analysis or multi-class text classification and it achieves state-of-the-art performance.

We used TFDistilBertForSequenceClassification to train this model and used BertForSequenceClassification annotator in Spark NLP 🚀 for prediction at scale!

Predicted Entities

negative, positive

Live Demo Open in Colab Download Copy S3 URI

How to use

document_assembler = DocumentAssembler() \
.setInputCol('text') \
.setOutputCol('document')

tokenizer = Tokenizer() \
.setInputCols(['document']) \
.setOutputCol('token')

sequenceClassifier = DistilBertForSequenceClassification \
.pretrained('distilbert_base_sequence_classifier_imdb', 'ur') \
.setInputCols(['token', 'document']) \
.setOutputCol('class') \
.setCaseSensitive(True) \
.setMaxSentenceLength(512)

pipeline = Pipeline(stages=[
document_assembler, 
tokenizer,
sequenceClassifier    
])

example = spark.createDataFrame([['یہ فلم واقعی اچھی تھی!']]).toDF("text")
result = pipeline.fit(example).transform(example)

val document_assembler = DocumentAssembler() 
.setInputCol("text") 
.setOutputCol("document")

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

val tokenClassifier = DistilBertForSequenceClassification.pretrained("distilbert_base_sequence_classifier_imdb", "ur")
.setInputCols("document", "token")
.setOutputCol("class")
.setCaseSensitive(true)
.setMaxSentenceLength(512)

val pipeline = new Pipeline().setStages(Array(document_assembler, tokenizer, sequenceClassifier))

val example = Seq("یہ فلم واقعی اچھی تھی!").toDS.toDF("text")

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

import nlu
nlu.load("ur.classify.distilbert_sequence.imdb").predict("""یہ فلم واقعی اچھی تھی!""")

Model Information

Data Source

https://huggingface.co/datasets/imdb_urdu_reviews

Benchmarking

precision    recall  f1-score   support

negative       0.94      0.96      0.95     24285
positive       0.96      0.95      0.95     25715

accuracy                           0.95     50000
macro avg       0.95      0.95      0.95     50000
weighted avg       0.95      0.95      0.95     50000