Finnish BERT Embeddings (Base Uncased)

Description

A version of Google’s BERT deep transfer learning model for Finnish. The model can be fine-tuned to achieve state-of-the-art results for various Finnish natural language processing tasks. FinBERT features a custom 50,000 wordpiece vocabulary that has much better coverage of Finnish words.

FinBERT has been pre-trained for 1 million steps on over 3 billion tokens (24B characters) of Finnish text drawn from news, online discussion, and internet crawls. By contrast, Multilingual BERT was trained on Wikipedia texts, where the Finnish Wikipedia text is approximately 3% of the amount used to train FinBERT.

These features allow FinBERT to outperform not only Multilingual BERT but also all previously proposed models when fine-tuned for Finnish natural language processing tasks.

Live Demo Open in Colab Download Copy S3 URI

How to use

...
embeddings = BertEmbeddings.pretrained("bert_finnish_uncased", "fi") \
      .setInputCols("sentence", "token") \
      .setOutputCol("embeddings")
nlp_pipeline = Pipeline(stages=[document_assembler, sentence_detector, tokenizer, embeddings])
pipeline_model = nlp_pipeline.fit(spark.createDataFrame([[""]]).toDF("text"))
result = pipeline_model.transform(spark.createDataFrame([['Rakastan NLP: tä']], ["text"]))

...
val embeddings = BertEmbeddings.pretrained("bert_finnish_uncased", "fi")
      .setInputCols("sentence", "token")
      .setOutputCol("embeddings")
val pipeline = new Pipeline().setStages(Array(document_assembler, sentence_detector, tokenizer, embeddings))
val data = Seq("Rakastan NLP: tä").toDF("text")
val result = pipeline.fit(data).transform(data)

import nlu

text = ["Rakastan NLP: tä"]
embeddings_df = nlu.load('fi.embed.bert.uncased.').predict(text, output_level='token')
embeddings_df

Results

	token	    fi_embed_bert_uncased__embeddings
		
      Rakastan  [-0.5126021504402161, -1.1741008758544922, 0.6...
 	NLP 	    [1.4763829708099365, -1.5427947044372559, 0.80...
 	: 	    [-0.2581554353237152, -0.5670831203460693, -1....
 	tä 	    [0.39770740270614624, -0.7221324443817139, 0.1...

Model Information

Data Source

The model is imported from https://github.com/TurkuNLP/FinBERT