|
gptkbp:instanceOf
|
gptkb:large_language_model
gptkb:model
|
|
gptkbp:architecture
|
gptkb:transformation
|
|
gptkbp:attentionHeads
|
12
|
|
gptkbp:availableOn
|
gptkb:Hugging_Face
gptkb:TensorFlow_Hub
|
|
gptkbp:basedOn
|
gptkb:BERT
|
|
gptkbp:category
|
gptkb:model
gptkb:NLP
gptkb:pretrained_model
deep learning
multilingual model
|
|
gptkbp:citation
|
gptkb:Devlin_et_al.,_2018
|
|
gptkbp:developedBy
|
gptkb:Google
|
|
gptkbp:embeddingSize
|
768
|
|
gptkbp:hasVariant
|
cased
uncased
|
|
gptkbp:input
|
gptkb:text
|
|
gptkbp:inputLengthLimit
|
512 tokens
|
|
gptkbp:layer
|
12
|
|
gptkbp:license
|
Apache 2.0
|
|
gptkbp:notTrainedOn
|
parallel corpora
|
|
gptkbp:openSource
|
true
|
|
gptkbp:output
|
contextual embeddings
|
|
gptkbp:parameter
|
110 million
340 million
|
|
gptkbp:predecessor
|
gptkb:BERT
|
|
gptkbp:releaseYear
|
2018
|
|
gptkbp:successor
|
gptkb:mBERT_(large)
gptkb:XLM-R
|
|
gptkbp:supportsLanguage
|
104 languages
|
|
gptkbp:tokenizerType
|
gptkb:WordPiece
|
|
gptkbp:trainer
|
gptkb:Wikipedia
masked language modeling
next sentence prediction
Wikipedia (104 languages)
|
|
gptkbp:usedFor
|
natural language processing
question answering
text classification
named entity recognition
|
|
gptkbp:usedIn
|
zero-shot learning
cross-lingual transfer learning
|
|
gptkbp:website
|
https://github.com/google-research/bert
|
|
gptkbp:bfsParent
|
gptkb:BERT
|
|
gptkbp:bfsLayer
|
7
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Multilingual BERT
|