English RoBERTa Embeddings (Sampling strategy 'sim select')


Pretrained RoBERTa Embeddings model, uploaded to Hugging Face, adapted and imported into Spark NLP. distilroberta-base-climate-s is a English model orginally trained by climatebert.

Sampling strategy s:As expressed in the author’s paper here, s is “sim select”, meaning 70% of the most similar sentences of one of the corpus was used, discarding the rest.


How to use

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

tokenizer = Tokenizer() \
    .setInputCols("document") \
embeddings = RoBertaEmbeddings.pretrained("roberta_embeddings_distilroberta_base_climate_s","en") \
    .setInputCols(["document", "token"]) \
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() 
val tokenizer = new Tokenizer() 

val embeddings = RoBertaEmbeddings.pretrained("roberta_embeddings_distilroberta_base_climate_s","en") 
    .setInputCols(Array("document", "token")) 

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: roberta_embeddings_distilroberta_base_climate_s
Compatibility: Spark NLP 3.4.2+
License: Open Source
Edition: Official
Input Labels: [sentence, token]
Output Labels: [bert]
Language: en
Size: 310.0 MB
Case sensitive: true


  • https://huggingface.co/climatebert/distilroberta-base-climate-s
  • https://arxiv.org/abs/2110.12010