English DistilBERT Embeddings (from Intel)

Description

Pretrained DistilBERT Embeddings model, uploaded to Hugging Face, adapted and imported into Spark NLP. distilbert-base-uncased-sparse-85-unstructured-pruneofa is a English model orginally trained by Intel.

Download

How to use

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

tokenizer = Tokenizer() \
    .setInputCols("document") \
    .setOutputCol("token")
  
embeddings = DistilBertEmbeddings.pretrained("distilbert_embeddings_distilbert_base_uncased_sparse_85_unstructured_pruneofa","en") \
    .setInputCols(["document", "token"]) \
    .setOutputCol("embeddings")
    
pipeline = Pipeline(stages=[documentAssembler, tokenizer, embeddings])

data = spark.createDataFrame([["I love Spark NLP"]]).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 = DistilBertEmbeddings.pretrained("distilbert_embeddings_distilbert_base_uncased_sparse_85_unstructured_pruneofa","en") 
    .setInputCols(Array("document", "token")) 
    .setOutputCol("embeddings")

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

val data = Seq("I love Spark NLP").toDF("text")

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

Model Information

Model Name: distilbert_embeddings_distilbert_base_uncased_sparse_85_unstructured_pruneofa
Compatibility: Spark NLP 3.4.2+
License: Open Source
Edition: Official
Input Labels: [sentence, token]
Output Labels: [bert]
Language: en
Size: 132.8 MB
Case sensitive: false

References

  • https://huggingface.co/Intel/distilbert-base-uncased-sparse-85-unstructured-pruneofa
  • https://arxiv.org/abs/2111.05754
  • https://github.com/IntelLabs/Model-Compression-Research-Package/tree/main/research/prune-once-for-all