facebook
/

bart-large-xsum

text2text-generation

Model card Files Files and versions

Bart model finetuned on xsum

docs: https://huggingface.co/transformers/model_doc/bart.html

finetuning: examples/seq2seq/ (as of Aug 20, 2020)

Metrics: ROUGE > 22 on xsum.

variants: search for distilbart

paper: https://arxiv.org/abs/1910.13461

Downloads last month: 4,728

Model tree for facebook/bart-large-xsum

Adapters

1 model

Finetunes

Quantizations

1 model

Spaces using facebook/bart-large-xsum 22

Paper for facebook/bart-large-xsum

BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension

Paper • 1910.13461 • Published Oct 29, 2019 • 6

Evaluation results

ROUGE-1 on cnn_dailymail
test set self-reported

25.270
ROUGE-2 on cnn_dailymail
test set self-reported

7.664
ROUGE-L on cnn_dailymail
test set self-reported

17.181
ROUGE-LSUM on cnn_dailymail
test set self-reported

21.793
loss on cnn_dailymail
test set self-reported

3.504
gen_len on cnn_dailymail
test set self-reported

27.446
ROUGE-1 on xsum
test set self-reported

45.453
ROUGE-2 on xsum
test set self-reported

22.346
ROUGE-L on xsum
test set self-reported

37.230
ROUGE-LSUM on xsum
test set self-reported

37.232