|
gptkbp:instance_of
|
gptkb:language
|
|
gptkbp:applies_to
|
self-supervised learning
|
|
gptkbp:architecture
|
gptkb:Transformers
|
|
gptkbp:developed_by
|
gptkb:Python
gptkb:Microsoft_Research
|
|
gptkbp:has
|
multiple variants
|
|
gptkbp:has_achieved
|
state-of-the-art performance
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
De BERTa v2
|
|
gptkbp:improves
|
gptkb:De_BERTa
|
|
gptkbp:is_available_on
|
gptkb:Hugging_Face_Model_Hub
|
|
gptkbp:is_based_on
|
BERT architecture
|
|
gptkbp:is_cited_in
|
NLP research
|
|
gptkbp:is_designed_for
|
improved understanding of context
|
|
gptkbp:is_documented_in
|
research papers
|
|
gptkbp:is_evaluated_by
|
gptkb:AX-bench
gptkb:GLUE_benchmark
gptkb:historical_memory
gptkb:SQu_AD
gptkb:MNLI
gptkb:Super_GLUE_benchmark
F1 score
accuracy
precision
WNLI
HANS
RTE
QQP
MRPC
Co LA
STSB
|
|
gptkbp:is_open_source
|
gptkb:true
|
|
gptkbp:is_optimized_for
|
performance on downstream tasks
|
|
gptkbp:is_part_of
|
De BERTa family
|
|
gptkbp:is_supported_by
|
gptkb:Tensor_Flow
gptkb:Py_Torch
|
|
gptkbp:is_trained_in
|
large text corpus
masked language modeling
next sentence prediction
|
|
gptkbp:is_used_for
|
question answering
sentiment analysis
text generation
text classification
named entity recognition
|
|
gptkbp:is_used_in
|
natural language processing tasks
|
|
gptkbp:outperforms
|
gptkb:BERT
gptkb:Ro_BERTa
|
|
gptkbp:performance
|
gptkb:GPT-3
gptkb:T5
gptkb:XLNet
ALBERT
ERNIE
NLP models
|
|
gptkbp:release_year
|
gptkb:2021
|
|
gptkbp:released_in
|
gptkb:2021
|
|
gptkbp:supports
|
multiple languages
|
|
gptkbp:uses
|
disentangled attention
|
|
gptkbp:bfsParent
|
gptkb:De_BERTa
|
|
gptkbp:bfsLayer
|
6
|