BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension
Paper • 1910.13461 • Published • 6
Summarization
Transformers
PyTorch
TensorFlow
JAX
Rust
English
bart
text2text-generation
Eval Results (legacy)
# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
tokenizer = AutoTokenizer.from_pretrained("facebook/bart-large-xsum")
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/bart-large-xsum")Quick Links
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
- Downloads last month
- 16,044
Model tree for facebook/bart-large-xsum
Spaces using facebook/bart-large-xsum 30
Paper for facebook/bart-large-xsum
Evaluation results
- ROUGE-1 on cnn_dailymailtest set self-reported25.270
- ROUGE-2 on cnn_dailymailtest set self-reported7.664
- ROUGE-L on cnn_dailymailtest set self-reported17.181
- ROUGE-LSUM on cnn_dailymailtest set self-reported21.793
- loss on cnn_dailymailtest set self-reported3.504
- gen_len on cnn_dailymailtest set self-reported27.446
- ROUGE-1 on xsumtest set self-reported45.453
- ROUGE-2 on xsumtest set self-reported22.346
- ROUGE-L on xsumtest set self-reported37.230
- ROUGE-LSUM on xsumtest set self-reported37.232
- loss on xsumtest set self-reported2.313
- gen_len on xsumtest set self-reported25.544
- ROUGE-1 on samsumself-reported24.785
- ROUGE-2 on samsumself-reported5.253
- ROUGE-L on samsumself-reported18.679
- ROUGE-LSUM on samsumself-reported20.629
# Use a pipeline as a high-level helper # Warning: Pipeline type "summarization" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("summarization", model="facebook/bart-large-xsum")