Korean Bert Embeddings (from deeq)


Pretrained Bert Embeddings model, uploaded to Hugging Face, adapted and imported into Spark NLP. dbert is a Korean model orginally trained by deeq.


How to use

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

tokenizer = Tokenizer() \
    .setInputCols("document") \
embeddings = BertEmbeddings.pretrained("bert_embeddings_dbert","ko") \
    .setInputCols(["document", "token"]) \
pipeline = Pipeline(stages=[documentAssembler, tokenizer, embeddings])

data = spark.createDataFrame([["나는 Spark NLP를 좋아합니다"]]).toDF("text")

result = pipeline.fit(data).transform(data)
val documentAssembler = new DocumentAssembler() 
val tokenizer = new Tokenizer() 

val embeddings = BertEmbeddings.pretrained("bert_embeddings_dbert","ko") 
    .setInputCols(Array("document", "token")) 

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

val data = Seq("나는 Spark NLP를 좋아합니다").toDF("text")

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

Model Information

Model Name: bert_embeddings_dbert
Compatibility: Spark NLP 3.4.2+
License: Open Source
Edition: Official
Input Labels: [sentence, token]
Output Labels: [bert]
Language: ko
Size: 424.1 MB
Case sensitive: true


  • https://huggingface.co/deeq/dbert