Papers of the day   All papers

LXMERT: Learning Cross-Modality Encoder Representations from Transformers

Comments

Mohit Bansal: Presenting LXMERT at @EMNLP2019 --> https://arxiv.org/abs/1908.07490 (prnc. 'leksmert'). Top3 in GQA & VQA challenges (May2019), Rank1 in VizWiz, & v.strong generalzn to NLVR2 (22% abs jump)! Awesome effort by @HaoTan5! CODE+MODELS all public: https://github.com/airsplay/lxmert; pls use+share! 1/2 https://t.co/WvxRirYGoB

2 replies, 124 likes


Mohit Bansal: Day3 talk2 #EMNLP2019 in 201A–C: @HaoTan5's talk on "LXMERT: Learning Cross-Modality Encoder Representations from Transformers" w. several new visualizations/analyses (+SotA on 3-4 vis-lang tasks) PS. I am hiring *POSTDOCS* & still around all day today+tmrw so pls chat/share!🙂 https://t.co/YiMJ3aMAYt

0 replies, 35 likes


Mohit Bansal (@🏡): @huggingface @HaoTan5 @avalmendoz @uncnlp Also thank u for your help @huggingface (@LysandreJik @qlhoest @Thom_Wolf)! Looking fwd to more community multimodal tasks built on top of @avalmendoz's efficient backend effort🤗 (ps. for new folks, original LXMERT details in @HaoTan5's #emnlp2019 paper https://arxiv.org/abs/1908.07490)

0 replies, 20 likes


Thomas Wolf: The work of @avalmendoz @HaoTan5 and @mohitban47 is crazy impressive, both on the research and the implementation levels. Diving deep in our datasets and models libs backend to unlock a whole new field of application, I’m amazed!

0 replies, 19 likes


Jin-Hwa Kim: New SotA on VQA (72.5), etc. -- LXMERT: Learning Cross-Modality Encoder Representations from Transformers (Tan & Bansal, EMNLP 2019) https://arxiv.org/abs/1908.07490 https://github.com/airsplay/lxmert

0 replies, 12 likes


William Wang: https://t.co/Cng1KgTMV0

1 replies, 12 likes


Verena Rieser: I just listened to @mohitban47 talking on this topic at the #NLPhighlights podcast hosted by @ai2_allennlp. Very interesting and lots to learn! https://soundcloud.com/nlp-highlights/107-multi-modal-transformers-with-hao-tan-and-mohit-bansal

2 replies, 11 likes


Mohit Bansal: Useful+interesting categorical and bias analysis (in the thread below) on #NLVR2 --> http://nlvr.ai! Details of LXMERT (pdf and code) --> https://arxiv.org/abs/1908.07490, https://github.com/airsplay/lxmert, https://twitter.com/mohitban47/status/1163978512094715904 @HaoTan5

0 replies, 4 likes


HotComputerScience: Most popular computer science paper of the day: "LXMERT: Learning Cross-Modality Encoder Representations from Transformers" https://hotcomputerscience.com/paper/lxmert-learning-cross-modality-encoder-representations-from-transformers https://twitter.com/mohitban47/status/1163978512094715904

0 replies, 4 likes


Frank Proctor: Fascinating talk by @HaoTan5 and @mohitban47 about their LEXMERT multimodal language+visual model, on @ai2_allennlp’s NLP highlights podcast https://bit.ly/2vdM9je Paper is here: https://arxiv.org/abs/1908.07490

0 replies, 4 likes


Mohit Bansal: @pdasigi @HaoTan5 Thanks again @PDasigi & @nlpMattG for hosting @HaoTan5+me for the fun discussion on cross-modal representation learning!🙂 All details about the paper again below: Pdf: https://arxiv.org/abs/1908.07490 Code: https://github.com/airsplay/lxmert Slides: http://www.cs.unc.edu/~airsplay/EMNLP_2019_LXMERT_slides.pdf https://twitter.com/mohitban47/status/1163978512094715904

0 replies, 3 likes


Mohit Bansal (@🏡): (please find the original details and info about LXMERT in @HaoTan5's #emnlp2019 paper at https://arxiv.org/abs/1908.07490)

0 replies, 2 likes


Rogue 🌻. Bigham: https://t.co/OwqpzS1Yjx

0 replies, 2 likes


Content

Found on Aug 21 2019 at https://arxiv.org/pdf/1908.07490.pdf

PDF content of a computer science paper: LXMERT: Learning Cross-Modality Encoder Representations from Transformers