BioBERT Embeddings (Clinical)


This model contains a pre-trained weights of ClinicalBERT for generic clinical text. This domain-specific model has performance improvements on 3/5 clinical NLP tasks andd establishing a new state-of-the-art on the MedNLI dataset. The details are described in the paper “Publicly Available Clinical BERT Embeddings”.


How to use

embeddings = BertEmbeddings.pretrained("biobert_clinical_base_cased", "en") \
      .setInputCols("sentence", "token") \
nlp_pipeline = Pipeline(stages=[document_assembler, sentence_detector, tokenizer, embeddings])
pipeline_model =[[""]]).toDF("text"))
result = pipeline_model.transform(spark.createDataFrame(pd.DataFrame({"text": ["I hate cancer"]})))

val embeddings = BertEmbeddings.pretrained("biobert_clinical_base_cased", "en")
      .setInputCols("sentence", "token")
val pipeline = new Pipeline().setStages(Array(document_assembler, sentence_detector, tokenizer, embeddings))
val result =["I hate cancer"].toDS.toDF("text")).transform(data)
import nlu

text = ["I hate cancer"]
embeddings_df = nlu.load('en.embed.biobert.clinical_base_cased').predict(text, output_level='token')


        token	en_embed_biobert_clinical_base_cased_embeddings
	I	[0.2206662893295288, 0.41324421763420105, -0.3...
	hate	[-0.19311018288135529, 0.6037888526916504, -0....
	cancer	[0.2895708680152893, 0.22499887645244598, -0.5...

Model Information

Model Name: biobert_clinical_base_cased
Type: embeddings
Compatibility: Spark NLP 2.6.2
License: Open Source
Edition: Official
Input Labels: [sentence, token]
Output Labels: [word_embeddings]
Language: [en]
Dimension: 768
Case sensitive: true

Data Source

The model is imported from