Japanese Bert Embeddings (from izumi-lab)

Description

Pretrained Bert Embeddings model, uploaded to Hugging Face, adapted and imported into Spark NLP. bert-small-japanese-fin is a Japanese model orginally trained by izumi-lab.

Download

How to use

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

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

data = spark.createDataFrame([["私はSpark NLPを愛しています"]]).toDF("text")

result = pipeline.fit(data).transform(data)
val documentAssembler = new DocumentAssembler() 
      .setInputCol("text") 
      .setOutputCol("document")
 
val tokenizer = new Tokenizer() 
    .setInputCols(Array("document"))
    .setOutputCol("token")

val embeddings = BertEmbeddings.pretrained("bert_embeddings_bert_small_japanese_fin","ja") 
    .setInputCols(Array("document", "token")) 
    .setOutputCol("embeddings")

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

val data = Seq("私はSpark NLPを愛しています").toDF("text")

val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("ja.embed.bert_small_japanese_fin").predict("""私はSpark NLPを愛しています""")

Model Information

Model Name: bert_embeddings_bert_small_japanese_fin
Compatibility: Spark NLP 3.4.2+
License: Open Source
Edition: Official
Input Labels: [sentence, token]
Output Labels: [bert]
Language: ja
Size: 68.3 MB
Case sensitive: true

References

  • https://huggingface.co/izumi-lab/bert-small-japanese-fin
  • https://github.com/google-research/bert
  • https://github.com/retarfi/language-pretraining/tree/v1.0
  • https://arxiv.org/abs/2003.10555
  • https://arxiv.org/abs/2003.10555
  • https://creativecommons.org/licenses/by-sa/4.0/