Papers of the day   All papers

On Extractive and Abstractive Neural Document Summarization with Transformer Language Models

Comments

Sep 10 2019 Miles Brundage

🧐 from: https://arxiv.org/abs/1909.03186 https://t.co/Ytv0uRAFE4
44 replies, 2897 likes


Sep 10 2019 Matthew Honnibal

Extremely strong flex here.
3 replies, 330 likes


Sep 10 2019 Jonathan Fly 👾

@Miles_Brundage This abstractive summarization paper abstract is a real roller coaster! https://arxiv.org/abs/1909.03186 https://t.co/MSKQ796Ulr
9 replies, 293 likes


Sep 10 2019 Sam Bowman

Cute/impressive. (From Subramanian et al.: https://arxiv.org/abs/1909.03186) https://t.co/3hFyxbMGVu
3 replies, 237 likes


Sep 11 2019 hardmaru

On Extractive and Abstractive Neural Document Summarization with Transformer Language Models 🧐https://arxiv.org/abs/1909.03186 https://t.co/4Mi9Vq5zWo
1 replies, 199 likes


Sep 10 2019 Peter Bloem

I'm usually critical of too much showmanship in science, but this is pretty cool.
1 replies, 157 likes


Sep 11 2019 Florian Jug

Wowww... mind officially blown!!! 🤯
5 replies, 111 likes


Sep 10 2019 Reiichiro Nakano

It's happening
0 replies, 86 likes


Sep 10 2019 Chelsea Parlett-Pelleriti

WHAT IF you never had to write an abstract again?! WHAT IF you could just give it a min/max length, and have it spit out the abstract. No more reformatting for different journal guidelines.🤯
3 replies, 79 likes


Sep 11 2019 Subbarao Kambhampati

@catherineols @sandsubramanian It looks like a big part of the "generated" abstract is almost verbatim passages from the end of the introduction (as @tchakra2 pointed out) May be the network "learned" that lazy abstracts involve repeating the end of a well written introduction? https://arxiv.org/pdf/1909.03186.pdf https://t.co/XzlYUJR6Qz
4 replies, 55 likes


Sep 10 2019 Jonathan Fly 👾

This abstractive summarization paper abstract is a real roller coaster! https://arxiv.org/abs/1909.03186 https://t.co/4rmgeKcmPG
0 replies, 32 likes


Sep 17 2019 Concerned bioinformatician

If paper abstracts are to be the next battleground in humans vs AI, then so be it. To clarify, Concerned Bioinformatician is a real life doge dog and not an artificial neural network of doge dogs.
3 replies, 31 likes


Sep 11 2019 Larry Hunter

Love the last line of the abstract https://arxiv.org/abs/1909.03186
2 replies, 29 likes


Sep 11 2019 Lê Nguyên Hoang (Science4All)

Note: The abstract above was not written by the authors, it was generated by one of the models presented in this paper. " AIs are now competing with humans at writing abstracts of scientific papers!! 😱😲😳 https://arxiv.org/abs/1909.03186
2 replies, 28 likes


Sep 10 2019 Helen Toner

In case it wasn't already clear that 2019 in machine learning is the Year of the Language Model, here's a paper about a language model that generates paper summaries where the summary of the paper was generated by the language model being summarized in the paper:
0 replies, 19 likes


Sep 13 2019 Nicolas Chapados

An NLP model that writes its own arXiv paper abstract?! Check out this great abstractive neural document summarization work by Element AI researchers Sandeep Subramanian, Raymond Li, Jonathan Pilault and Christopher Pal: https://arxiv.org/abs/1909.03186
0 replies, 15 likes


Sep 11 2019 Tim Miller

Clever abstract!
2 replies, 15 likes


Sep 10 2019 Neil Tenenholtz

💯% should have included the algo name in the list of authors.
2 replies, 12 likes


Sep 10 2019 nic lane

Paper about text summarization writes it’s own abstract... 😎
0 replies, 11 likes


Sep 11 2019 Emanuel

Now that's how you write an abstract! 🤖
0 replies, 10 likes


Sep 10 2019 Chase Roberts

If you haven't already, just read the abstract of this paper. https://arxiv.org/pdf/1909.03186.pdf
4 replies, 10 likes


Sep 11 2019 vladimir.alexeev

This re-tweet about is AI generated abstractive summarization paper was written by a human being.
1 replies, 7 likes


Sep 10 2019 David Chen

oh yes! One step closer to having an instant paper generator
1 replies, 5 likes


Sep 10 2019 Walter De Brouwer

The abstract on the article about abstract summarization was written by one of the abstract summarizer models!😊
1 replies, 5 likes


Sep 11 2019 Natalia Díaz Rodríguez

I was fantasizing with this idea for long ❤️
0 replies, 5 likes


Sep 10 2019 Jorge Rotter

🤯
0 replies, 4 likes


Sep 10 2019 Daniele Grattarola

Science done right
0 replies, 4 likes


Sep 10 2019 Ibrahim Sharaf

Oh oh!
1 replies, 4 likes


Sep 10 2019 Mehrdad Hosseinzadeh

How to pitch your idea in the best possible way!
0 replies, 4 likes


Sep 11 2019 Daisuke Okanohara

Finally, we see the paper whose abstract is written by the proposed method. Extract important sentences by pointer network and train transformer LM for the concatenated document of the introduction, extracted sentences, abstract, and the rest of the paper. https://arxiv.org/abs/1909.03186
0 replies, 4 likes


Sep 10 2019 Felix Busch

Next up: The paper below was not written by humans, it was generated by one of the models presented in the abstract.
0 replies, 3 likes


Sep 10 2019 Eric Xingdi Yuan

wow!
0 replies, 3 likes


Sep 10 2019 Iván Rivera

😮
1 replies, 3 likes


Sep 12 2019 Granta Innovation

Document summarisation is an exciting frontier in #AI research, so it's fascinating to see these results from a team at Element AI/U. Montreal h/t @miguelgfierro . This paper's abstract was written by an AI system: https://arxiv.org/abs/1909.03186 https://t.co/dmLEI2V8KZ
0 replies, 3 likes


Sep 13 2019 Thomas Scialom

#NLProc papers: Abstractive model for long document summarization. https://arxiv.org/abs/1909.03186 authors: @Sandeepsub et al. https://t.co/S7FRj6XQLv
1 replies, 3 likes


Sep 10 2019 Lewis Shepherd

Check out the footnote
0 replies, 3 likes


Sep 10 2019 Nick Shchetko

Nice! This abstract in a paper on AI-generated abstracts was generated by an abstract generator model from the paper https://arxiv.org/pdf/1909.03186.pdf https://t.co/l3jZGgP3wA
0 replies, 2 likes


Nov 02 2019 Stephen Pimentel

"The abstract above was not written by the authors, it was generated by one of the models presented in this paper." On Extractive and Abstractive Neural Document Summarization with Transformer Language Models https://arxiv.org/abs/1909.03186
0 replies, 2 likes


Sep 13 2019 Aaron Schumacher

Can you really trust a document summarization paper where the authors write the paper's abstract _by hand_? https://arxiv.org/abs/1909.03186 https://t.co/uHAWM5h1QK
1 replies, 1 likes


Sep 10 2019 Simon J Greenhill

On Extractive and Abstractive Neural Document Summarization with Transformer Language Models: "Note: The abstract above was not written by the authors, it was generated by one of the models presented in this paper." https://arxiv.org/abs/1909.03186
0 replies, 1 likes


Sep 11 2019 Rahul Devanarayanan

YOOOOOOOOO 😤😤😤🤯🤯🤯😩😩😩💦💦💦
0 replies, 1 likes


Sep 10 2019 Jesper Dramsch

Whelp.
1 replies, 1 likes


Sep 10 2019 Daddy Warpig

WRITERS, THE COMPUTERS ARE COMING FOR YOUR JOBS.
0 replies, 1 likes


Sep 16 2019 Frank Norman

Brilliant! Turing would’ve been proud.
0 replies, 1 likes


Sep 10 2019 John Austin

NO
1 replies, 0 likes


Content