English RobertaForMaskedLM Base Cased model

Description

Pretrained RobertaForMaskedLM model, adapted from Hugging Face and curated to provide scalability and production-readiness using Spark NLP. distilroberta-base is a English model originally trained by HuggingFace.

Live Demo Open in Colab Download Copy S3 URI

How to use

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

tokenizer = Tokenizer() \
    .setInputCols("document") \
    .setOutputCol("token")

roberta_loaded = RoBertaEmbeddings.pretrained("roberta_embeddings_distil_base","en") \
    .setInputCols(["document", "token"]) \
    .setOutputCol("embeddings") \
    .setCaseSensitive(True)

pipeline = Pipeline(stages=[documentAssembler, tokenizer, roberta_loaded])

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("document")
    .setOutputCol("token")

val roberta_loaded = RoBertaEmbeddings.pretrained("roberta_embeddings_distil_base","en")
    .setInputCols(Array("document", "token"))
    .setOutputCol("embeddings")
    .setCaseSensitive(true)

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

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

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

import nlu
nlu.load("en.embed.roberta.distilled_base").predict("""I love Spark NLP""")

Model Information

References

• https://huggingface.co/distilroberta-base
• https://arxiv.org/abs/1910.01108
• https://aclanthology.org/2021.acl-long.330.pdf
• https://dl.acm.org/doi/pdf/10.1145/3442188.3445922
• https://skylion007.github.io/OpenWebTextCorpus/
• https://mlco2.github.io/impact#compute
• https://arxiv.org/abs/1910.09700