Description
Longformer Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks.
longformer_large_token_classifier_conll03 is a fine-tuned Longformer model that is ready to use for Named Entity Recognition and achieves state-of-the-art performance for the NER task. This model has been trained to recognize four types of entities: location (LOC), organizations (ORG), person (PER), and Miscellaneous (MISC).
We used TFLongformerForTokenClassification to train this model and used LongformerForTokenClassification annotator in Spark NLP 🚀 for prediction at scale!
Predicted Entities
PER, LOC, ORG, MISC
How to use
document_assembler = DocumentAssembler() \
.setInputCol('text') \
.setOutputCol('document')
tokenizer = Tokenizer() \
.setInputCols(['document']) \
.setOutputCol('token')
tokenClassifier = LongformerForTokenClassification \
.pretrained('longformer_large_token_classifier_conll03', 'en') \
.setInputCols(['token', 'document']) \
.setOutputCol('ner') \
.setCaseSensitive(True) \
.setMaxSentenceLength(512)
# since output column is IOB/IOB2 style, NerConverter can extract entities
ner_converter = NerConverter() \
.setInputCols(['document', 'token', 'ner']) \
.setOutputCol('entities')
pipeline = Pipeline(stages=[
document_assembler,
tokenizer,
tokenClassifier,
ner_converter
])
example = spark.createDataFrame([['My name is John!']]).toDF("text")
result = pipeline.fit(example).transform(example)
val document_assembler = DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val tokenizer = Tokenizer()
.setInputCols("document")
.setOutputCol("token")
val tokenClassifier = LongformerForTokenClassification.pretrained("longformer_large_token_classifier_conll03", "en")
.setInputCols("document", "token")
.setOutputCol("ner")
.setCaseSensitive(true)
.setMaxSentenceLength(512)
// since output column is IOB/IOB2 style, NerConverter can extract entities
val ner_converter = NerConverter()
.setInputCols("document", "token", "ner")
.setOutputCol("entities")
val pipeline = new Pipeline().setStages(Array(document_assembler, tokenizer, tokenClassifier, ner_converter))
val example = Seq.empty["My name is John!"].toDS.toDF("text")
val result = pipeline.fit(example).transform(example)
import nlu
nlu.load("en.ner.longformer.conll.large").predict("""My name is John!""")
Results
+------------------------------------------------------------------------------------+
|result |
+------------------------------------------------------------------------------------+
|[B-PER, I-PER, O, O, O, B-LOC, O, O, O, B-LOC, O, O, O, O, B-PER, O, O, O, O, B-LOC]|
+------------------------------------------------------------------------------------+
Model Information
| Model Name: | longformer_large_token_classifier_conll03 |
| Compatibility: | Spark NLP 3.3.0+ |
| License: | Open Source |
| Edition: | Official |
| Input Labels: | [token, document] |
| Output Labels: | [ner] |
| Language: | en |
| Case sensitive: | true |
| Max sentense length: | 4096 |
Data Source
https://www.clips.uantwerpen.be/conll2003/ner/
Benchmarking
precision recall f1-score support
B-LOC 0.97 0.92 0.94 1837
B-MISC 0.93 0.88 0.90 922
B-ORG 0.91 0.92 0.92 1341
B-PER 0.96 0.98 0.97 1842
I-LOC 0.91 0.92 0.92 257
I-MISC 0.89 0.82 0.85 346
I-ORG 0.90 0.94 0.92 751
I-PER 0.98 0.98 0.98 1307
O 1.00 1.00 1.00 42759
accuracy 0.99 51362
macro avg 0.94 0.93 0.93 51362
weighted avg 0.99 0.99 0.99 51362
processed 51362 tokens with 5942 phrases; found: 5900 phrases; correct: 5504.
accuracy: 93.75%; (non-O)
accuracy: 98.70%; precision: 93.29%; recall: 92.63%; FB1: 92.96
LOC: precision: 96.19%; recall: 92.00%; FB1: 94.05 1757
MISC: precision: 90.28%; recall: 86.66%; FB1: 88.43 885
ORG: precision: 88.81%; recall: 90.53%; FB1: 89.66 1367
PER: precision: 95.24%; recall: 97.77%; FB1: 96.49 1891