Kenton Lee @kentonctlee
Joined January 2015-
Tweets31
-
Followers853
-
Following112
-
Likes56
Excited to present Pix2Struct! It's a general-purpose pixel-to-text model that can be finetuned on tasks with visually-situated language, such as UIs, charts, figures, tables, documents, etc. -- arxiv.org/abs/2210.03347 (1/4)
New from Google Research! Retrieval tasks are quite different -- Few-shot adaptation is important! With 8 annotated examples, Promptagator🐊 can figure out the task and make dual encoders outperform all prior MSMARCO trained retrievers in experiments arxiv.org/abs/2209.11755 🧵1/n
New from Google Research: arxiv.org/abs/2106.16171 To build multilingual NLP systems, a successful recipe is to pre-train on a multilingual corpus, and then fine-tune on labeled data in a single transfer language -- usually English. But is English best?
New from Google Research! arxiv.org/abs/2102.01335 Show examples from the distribution you want, and our example extrapolator (Ex2) generates new examples from the same distribution. We use Ex2 for data augmentation, improving over SOTA methods on multiple NLP benchmarks! (1/3)
Excited to share our new EMNLP paper, “CapWAP: Captioning with a Purpose”! We take a fresh perspective on image captioning by giving it an explicit purpose. In CapWAP, models are trained and evaluated with respect to information that users care about. arxiv.org/abs/2011.04264
Knowledge is not uniformly distributed across languages so an ideal open-retrieval QA model should search and retrieve multilingual resources. We propose a new task 𝐗𝐎𝐑 𝐐𝐀 with a new dataset 𝐗𝐎𝐑-𝐓𝐲𝐃𝐢 QA(40k newly annotated Qs in 7 languages) nlp.cs.washington.edu/xorqa/👇
We're doing a live online Q&A on REALM: Retrieval Augmented Language Model Pre-training at #ICML2020: icml.cc/virtual/2020/p… (Tue Jul 14, 9 AM PDT and again at 8 PM PDT).
Announcing the EfficientQA competition and #NeurIPS2020 workshop, a collaborative effort with @Princeton and @UW that challenges developers to create end-to-end open-domain question answering systems that are small, yet robust. Learn all about it ↓ goo.gle/2AVm3Vg
New paper on representing texts by contextualizing them jointly with textual encyclopedic knowledge (TEK) retrieved dynamically from multiple documents in Wikipedia. Joint work with @kentonctlee, Yi Luan & @toutanova from my internship at @GoogleAI. arxiv.org/abs/2004.12006 (1/4)
Our BERT miniatures were pre-trained directly with MLM loss. They are competitive to more elaborate pre-training strategies involving MLM distillation (arxiv.org/abs/1908.08962). Our models can be fine-tuned for downstream tasks via standard training or end-task distillation.
Efficient BERT models from Google Research, now available at github.com/google-researc…! We hope our 24 BERT models with fewer layers and/or hidden sizes will enable research in resource-constrained institutions and encourage building more compact models. arxiv.org/abs/1908.08962
@kentonctlee @IcePasupat @mchang21 Instead of compressing massive textual knowledge into billions of uninterpretable parameters, REALM exposes the provenance for its representations and decision making. Corollary: we can control what REALM “knows” by hot-swapping the corpus.
New from Google Research! REALM: realm.page.link/paper We pretrain an LM that sparsely attends over all of Wikipedia as extra context. We backprop through a latent retrieval step on 13M docs. Yields new SOTA results for open domain QA, breaking 40 on NaturalQuestions-Open!
Shoutout to @JonClarkSeattle from our team in Seattle, who led this project (in collaboration with many other stellar researchers from our group in NYC); it is a rigorously collected dataset for 11 diverse languages for information-seeking QA. Please try it!
Shoutout to @JonClarkSeattle from our team in Seattle, who led this project (in collaboration with many other stellar researchers from our group in NYC); it is a rigorously collected dataset for 11 diverse languages for information-seeking QA. Please try it!
What goes on in the mind of BERT? Using an interactive visualization tool, we uncovered some surprisingly intuitive patterns. Article: medium.com/@JesseVig/deco… Colab: colab.research.google.com/drive/1vlOJ1lh… Github: github.com/jessevig/bertv… Play with it and share what you find! #nlproc
Code and pretrained weights for BERT are out now. Includes scripts to reproduce results. BERT-Base can be fine-tuned on a standard GPU; for BERT-Large, a Cloud TPU is required (as max batch size for 12-16 GB is too small). github.com/google-researc…
New preprint from our team in Seattle. State of the art on everything! arxiv.org/abs/1810.04805
A new era of NLP has just begun a few days ago: large pretraining models (Transformer 24 layers, 1024 dim, 16 heads) + massive compute is all you need. BERT from @GoogleAI: SOTA results on everything arxiv.org/abs/1810.04805. Results on SQuAD are just mind-blowing. Fun time ahead!
the paper behind BERT is now online: arxiv.org/abs/1810.04805 BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding Jacob Devlin, Ming-Wei Chang, Kenton Lee, Kristina Toutanova arxiv.org/abs/1810.04805
(((ل()(ل() 'yoav))).. @yoavgo
46K Followers 2K FollowingAkari Asai @AkariAsai
11K Followers 650 Following Ph.D. student @uwcse & @uwnlp. NLP. IBM Ph.D. fellow (2022-2023). Meta student researcher (2023-) . ☕️ 🐕 🏃♀️🧗♀️🍳Percy Liang @percyliang
49K Followers 408 Following Associate Professor in computer science @Stanford @StanfordHAI @StanfordCRFM @StanfordAILab @stanfordnlp | cofounder @togethercompute | PianistKyunghyun Cho @kchonyc
61K Followers 2K Following a combination of a mediocre scientist, a mediocre manager, a mediocre advisor & a mediocre PC at @nyuniversity (@CILVRatNYU) & @genentech (@PrescientDesign).Yoav Artzi @yoavartzi
13K Followers 163 Following Research/prof @cs_cornell + @cornell_tech🚡 / https://t.co/9YnWry7yHs / https://t.co/3VmRSyYm2d / asso. faculty director @arxiv / building https://t.co/f9QkzO5kaCSam Bowman @sleepinyourhat
35K Followers 3K Following AI alignment + LLMs at NYU & Anthropic. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.Jacob Andreas @jacobandreas
14K Followers 958 Following Teaching computers to read. Assoc. prof @MITEECS / @MIT_CSAIL (he/him). https://t.co/5kCnXHjtlY https://t.co/2A3qF5vdJwOfir Press 🖋 @OfirPress
10K Followers 3K Following I build tough benchmarks for LMs and then I get the LMs to solve them. Postdoc @Princeton. PhD from @nlpnoah @UW. Ex-visiting researcher @MetaAI & @MosaicML.Dipanjan Das @dipanjand
4K Followers 308 Following Senior Director of Research at @GoogleDeepmind. Working on improving the factuality of LLM generated content.Graham Neubig @gneubig
31K Followers 588 Following Associate professor at CMU, studying natural language processing and machine learning.Tim Dettmers @Tim_Dettmers
29K Followers 821 Following PhD Student at @UW. I blog about deep learning and PhD life at https://t.co/Y78KDJJFE7.Wenhu Chen @WenhuChen
11K Followers 520 Following AI researcher @UWaterloo @GoogleAI @VectorInst. Interested in natural language processing, diffusion models. I direct TIGER-Lab at UWaterloo.Delip Rao e/σ @deliprao
46K Followers 5K Following Busy inventing the shipwreck. @Penn. Past: @johnshopkins, @UCSC, @Amazon, @Twitter ||Art: #NLProc, Vision, Speech, #DeepLearning || Life: 道元, improv, running 🌈Victor Zhong @hllo_wrld
4K Followers 450 Following ML+NLP assistant prof @UWCheritonCS. Formerly @MSFTResearch @MetaAI, @SFResearch via @MetamindIO, @uwnlp, @StanfordNLP, @eceuoft.Gabriel Ilharco @gabriel_ilharco
4K Followers 1K Following Building cool things @xAI. Prev. PhD at UW, Google AIGreg Durrett @gregd_nlp
6K Followers 752 Following CS professor at UT Austin. I do NLP most of the time. he/himOri Ram @ori__ram
765 Followers 386 Following Research Scientist @GoogleAI, working on #NLProc. Previously: PhD from @TelAvivUni, Research Scientist @AI21LabsMr. Jack Tung @MrJackTung
207 Followers 3K FollowingMarshall McLuhan @MarshallMcLuan
12 Followers 136 FollowingPúdrete Flanders @_CorvenDallas_
79 Followers 2K Followingcamenduru @camenduru
15K Followers 4K Following ML & Computer Engineer, Game Designer. #OpenSource ❤ #UE ❤ #Jupyter ❤ #AI #ML #StableDiffusion #LLM #NeRF #GaussianSplatting #T2V https://t.co/8MMNbygz1PPensé FFun @inftyCategory
98 Followers 6K FollowingArkil @arkil_patel
758 Followers 828 Following PhD Student at Mila (@Mila_Quebec) and McGill (@mcgillu) | Research in ML/NLP | Prev @allen_ai @MSFTResearch | alum @bitspilaniindiaRong Ching Chang @AnnCC12
690 Followers 5K Following Fascinated by ML, LLMs, GNN, Multimodal models in social media. Ph.D. student @ucdavisGokul Swamy @g_k_swamy
2K Followers 1K Following phd candidate @CMU_Robotics. ms @berkeley_ai. summers @GoogleAI, @msftresearch, @aurora_inno, @nvidia, @spacex. no model is an island.Ekin Akyürek @akyurekekin
2K Followers 726 Following graduate student in computer science @MITEECS/@MIT_CSAILMogan @IAmMogan
243 Followers 290 Following President @Fullstack (acq. by Zovio), @gracehopperfsa. Previously @Invitemedia (acq. by @Google), #tigerbeer. Die-hard @LFC fan. Tweets my ownVicky Zayats @ZayatsVicky
7 Followers 87 FollowingMaddison Mckenzie @MaddisonMc32252
2 Followers 13 FollowingBárbara Zita Peters .. @zpcbarbara
84 Followers 401 Following @[email protected] PhD student at @saezlab 💻🧬Audrey Peck @PeckAudrey42690
0 Followers 10 FollowingLuke Waltz @lwaltz12
24 Followers 342 FollowingSamantha Cox @SamanthaCo40420
0 Followers 11 FollowingIsabel Hubbard @hubbard_is48570
0 Followers 11 FollowingEvan Reynolds @EvanReynol25786
0 Followers 11 Followingturing — e/acc @linus_turing
612 Followers 5K FollowingLincoln Downs @DownsLinco93088
6 Followers 11 FollowingVoyage AI @Voyage_AI_
2K Followers 164 Following Building embedding/vectorization models, customized for your domain and company, for better retrieval quality https://t.co/MEAhTpBQqdTengyu Ma @tengyuma
26K Followers 512 Following Assistant professor at Stanford; Co-founder of Voyage AI (https://t.co/wpIITHLgF0) ; Working on ML, DL, RL, LLMs, and their theory.Safaa Rojas @RojasSafaa93325
1 Followers 11 FollowingIsabel Adams @IsabelAdam71613
0 Followers 11 FollowingDawson Rubio @DawsonRubi69936
0 Followers 11 FollowingRonan Galvan @RonanGalva5072
0 Followers 11 FollowingAbdur Gates @GatesAbdur95373
1 Followers 14 FollowingYuxuan Sun @jamesja69137043
77 Followers 175 Following Currently working on pathology #LMM and #Medical image analysis.Aoife Parks @AoifeParks73203
3 Followers 16 FollowingKarel D’Oosterlinck @KarelDoostrlnck
2K Followers 593 Following Interpretable AI, RAG, Biomedical NLP. Intern @ContextualAI, PhD student @ugent, visitor @stanfordnlp. Instigator of hikes.Richard Diehl Martine.. @richarddm1
115 Followers 992 Following CS PhD at University of Cambridge. Previously Applied Scientist @Amazon, MS/BS @Stanford.Jaeyoung Lee @lee__jaeyoung
22 Followers 203 FollowingCheungZee @CheungZee
19 Followers 24 Following touch me https://t.co/RQkoB67jdy http://t.co/r46PgSCPKhChaitanya Malaviya @cmalaviya11
99 Followers 121 Following PhD student at UPenn | currently @GoogleDeepMindKai Zhang @DrogoKhal4
1K Followers 641 Following PhD student @osunlp. Ex @MSFTResearch and @GoogleDeepMind.Chloe Valdez @ChloeValde76655
3 Followers 16 FollowingXuxing Chen @XuxingChen3
283 Followers 3K FollowingMuni Kumar @sssvmkumar
102 Followers 4K FollowingChristopher Harris @Christo51860350
2 Followers 11 Followingjohn kreesereb @JKreesereb
3 Followers 63 FollowingAref Jafari @ArefJafariii
4 Followers 46 FollowingMinji Yoon @MinjiYoon90
2K Followers 304 Following Building personal AI @MicrosoftAI. Past: @InflectionAI, PhD @SCSatCMU.Guanzhi Wang @guanzhi_wang
1K Followers 731 Following @Caltech CS PhD. @NVIDIA GEAR Lab Research Intern. LLM, Robotics, and Gaming AI. Voyager, MineDojo, Eureka. Ex: @StanfordAILabHarpreet Singh @harpreet_utd
2 Followers 714 FollowingRaphael Schumann @RaphiRaph_
368 Followers 1K Following Natural Language Processing PhD Student @ Heidelberg University.Rick Lamers @RickLamers
2K Followers 867 Following 👨💻 AI Research & Engineering @GroqInc. I publish a weekly update about LLM Engineering on Substack, it’s free. Opinions are my own.(((ل()(ل() 'yoav))).. @yoavgo
46K Followers 2K FollowingAkari Asai @AkariAsai
11K Followers 650 Following Ph.D. student @uwcse & @uwnlp. NLP. IBM Ph.D. fellow (2022-2023). Meta student researcher (2023-) . ☕️ 🐕 🏃♀️🧗♀️🍳Percy Liang @percyliang
49K Followers 408 Following Associate Professor in computer science @Stanford @StanfordHAI @StanfordCRFM @StanfordAILab @stanfordnlp | cofounder @togethercompute | PianistKyunghyun Cho @kchonyc
61K Followers 2K Following a combination of a mediocre scientist, a mediocre manager, a mediocre advisor & a mediocre PC at @nyuniversity (@CILVRatNYU) & @genentech (@PrescientDesign).Yoav Artzi @yoavartzi
13K Followers 163 Following Research/prof @cs_cornell + @cornell_tech🚡 / https://t.co/9YnWry7yHs / https://t.co/3VmRSyYm2d / asso. faculty director @arxiv / building https://t.co/f9QkzO5kaCSasha Rush @srush_nlp
52K Followers 464 Following Professor, Programmer in NYC. Cornell Tech, Hugging Face 🤗 https://t.co/cZl0wTfqGzSam Bowman @sleepinyourhat
35K Followers 3K Following AI alignment + LLMs at NYU & Anthropic. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.Mike Lewis @ml_perception
6K Followers 227 Following Llama3 pre-training lead. Partially to blame for things like the Cicero Diplomacy bot, BART, RoBERTa, kNN-LM, top-k sampling & Deal Or No Deal.Jacob Andreas @jacobandreas
14K Followers 958 Following Teaching computers to read. Assoc. prof @MITEECS / @MIT_CSAIL (he/him). https://t.co/5kCnXHjtlY https://t.co/2A3qF5vdJwChristopher Manning @chrmanning
127K Followers 116 Following Director, @StanfordAILab. Assoc. Director, @StanfordHAI. Founder, @stanfordnlp. Prof. CS & Linguistics, @Stanford. IP @aixventureshq. 🇦🇺 Do #NLProc & #AI. 👋Dipanjan Das @dipanjand
4K Followers 308 Following Senior Director of Research at @GoogleDeepmind. Working on improving the factuality of LLM generated content.Graham Neubig @gneubig
31K Followers 588 Following Associate professor at CMU, studying natural language processing and machine learning.Tim Dettmers @Tim_Dettmers
29K Followers 821 Following PhD Student at @UW. I blog about deep learning and PhD life at https://t.co/Y78KDJJFE7.Greg Durrett @gregd_nlp
6K Followers 752 Following CS professor at UT Austin. I do NLP most of the time. he/himColin Raffel @colinraffel
30K Followers 655 Following nonbayesian parameterics, sweet lessons, and random birds. Friend of @srush_nlpUri Shaham @Uri_Shaham
365 Followers 242 Following PhD candidate at Tel-Aviv University. Research intern at Google Research.Victoria X Lin @VictoriaLinML
3K Followers 761 Following Research Scientist @AIatMeta Foundational AI Research • ex-@SFResearch • PhD @uwcse 📜 https://t.co/j6QTac5q0rFelix Hill @FelixHill84
9K Followers 777 Following Research Scientist, Deepmind I try to think hard about everything I tweet, esp on 90s football and 80s music None of my opinions are really someone else'sVicky Zayats @ZayatsVicky
7 Followers 87 FollowingYang Chen @ychenNLP
664 Followers 438 FollowingPanupong Pasupat @IcePasupat
263 Followers 130 Following A research scientist working on natural language processing.Patrick Lewis @PSH_Lewis
4K Followers 656 Following London-based AI/NLP Research Scientist. I co-lead the RAG & tool use team at Cohere w/ @s_hofstaetter. Previous Fundamental AI Research at Meta AI, FAIR, UCL AIDerek Chen @derekchen14
574 Followers 271 Following Research scientist in conversational AI. Building @soleda_ai through scalable data generation. Prev: @columbianlp, @asapp, @UW, @stanfordnlp, @UCBerkeleyDanny To Eun Kim @TEKnologyy
327 Followers 930 Following PhD-ing @LTIatCMU working with @841io. MEng @ai_ucl | NLP & IRAakanksha Chowdhery @achowdhery
7K Followers 3K Following LLMs @ Google DeepMind :: PaLM, Gemini // Previously @MSFTResearch, @Stanford, @Princeton // views my own and subject to changeThibault Sellam @ThiboIbo
345 Followers 268 FollowingMaithra Raghu @maithra_raghu
17K Followers 476 Following Cofounder and CEO @Samaya_AI. Formerly Research Scientist Google Brain (@GoogleAI), PhD in ML @Cornell.Hexiang (Frank) Hu @Hexiang_Hu
506 Followers 401 Following Research Scientist @GoogleDeepmind | Vision & Language | GeminiXingyao Wang @xingyaow_
894 Followers 938 Following PhD student @IllinoisCS | BS @UMichCSE ('22) | Ex Intern @GoogleAI @Microsoft | Natural Language Processing | OpenDevin Core ContributorIan Magnusson @IanMagnusson
250 Followers 294 Following Predoctoral Young Investigator on AllenNLP at @allen_ai. Working on domain adaptation, reproducibility, and evaluation in NLP.Spandana Gella @gspandana
774 Followers 416 Following Mom. Researcher in #NLProc at Amazon AI. Former PhD student at @EdinburghNLP, Intern @MetaAI, MSR.UMassNLP @UMass_NLP
1K Followers 383 Following Natural language processing group at UMass Amherst @umasscs. Led by @thompson_laure @MohitIyyer @brendan642 @andrewmccallum #nlprocMaxwell Forbes @maxforbes
386 Followers 85 Following developer of https://t.co/vmZvh3tPqS, prev: travel, phd @uwcse (NLP), eng @google. purveyor of graphics / photos / essays. mediocre generalistJordan Boyd-Graber @boydgraber
4K Followers 2K Following Trivia Nerd, NLPer, Dad, Colorado native in Maryland exile Working on QA, negotiating/cooperating bots, ML explanations Exemplar for absent-minded professorBhuwan Dhingra @bhuwandhingra
928 Followers 290 Following Natural Language Processing / Machine Learning research. Assistant Professor @dukecompsci, @duke_nlp; Research Scientist @googleaiNando de Freitas 🏳.. @NandoDF
97K Followers 659 Following I research intelligence to understand it and to harness it wisely. Part of AlphaGo tuning, AlphaCode, learning to learn, Lyria, Imagen2, Gato, rGemmaUrvashi Khandelwal @ukhndlwl
2K Followers 611 Following Research Scientist @GoogleDeepMind, Stanford CS PhD @stanfordnlpChenhao Tan @ChenhaoTan
4K Followers 902 Following Assistant professor @UChicagoCS @UChicago. Working on human-centered AI, NLP, CSS at @ChicagoHAI, also part of @ChicagoNLP. DM for Postdoc/PhD opportunities.Adam Fisch @adamjfisch
1K Followers 246 Following Research Scientist @ Google DeepMind | Formerly: PhD @ MIT EECS.Thibault Févry @iwontbecreative
756 Followers 3K Following Researcher Point72, prev NLP Research @Google @Benevolent_ai, @NYUDataScience.Alexis Conneau @alex_conneau
24K Followers 113 Following Audio AGI Research Lead @OpenAI - GPT-Next - Past: XLM, Unsupervised ASR, Unsupervised MT, Wav2vec 2.0/XLSR, MUSE, Unsupervised cross-lingual transferMichael Roth @microth
1K Followers 49 Following Computational linguist, organizer of LSDSem/UnImplicit workshops, leading a research group on understanding misunderstandings.Marjan Ghazvininejad @gh_marjan
2K Followers 340 Following Research Scientist at FacebooK AI Research (FAIR), #NLProcElizabeth Clark @eaclark07
1K Followers 252 Following Doing NLP research at @GoogleAI. PhD from @uwcse.Ankur Parikh @ank_parikh
3K Followers 3K Following Staff Research Scientist at Google DeepMind. Former adjunct assistant prof at @NYU_Courant. PhD at @mldcmu. ML for Bio/Chem (Prev. NLP). All opinions my own.Kyle Lo @kylelostat
2K Followers 1K Following #nlproc #hci leading data research @allen_ai, he/him, bluesky https://t.co/5Hm9cx3mC1Jesse Dodge @JesseDodge
3K Followers 2K Following Senior Research Scientist at AI2 @ai2_allennlp. Responsibly open work on the science of AI and AI for science. Environmental impact of AI. he/him 🏳️🌈Ves Stoyanov @vesko_st
2K Followers 550 Following Head of AI at @magicaltome. Ex-Language Researcher at @FacebookAI. Large LMs and multilingual NLP. @JHUCLSP and @Cornell alumn. https://t.co/WTSCasqDI6Jeff Dean (@🏡) @JeffDean
296K Followers 6K Following Chief Scientist, Google DeepMind and Google Research. Co-designer/implementor of things like @TensorFlow, MapReduce, Bigtable, Spanner, Gemini .. (he/him)📍🧵🚨 QA on plots & charts is a complex task requiring sophisticated reasoning - our visual language models struggle with this. LLMs are super strong reasoners - but they only work for text. What do we do? We translate plots & charts to text so LLM can understand!
MATCHA : Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering Outperforms SotA by up to 20% on PlotQA and ChartQA - Transfers well to domains like screenshot, diagrams, and document figures arxiv.org/abs/2212.09662
Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding Pretrains a model for VL understanding by pairing a screenshot of a website and the simplified version of the HTML. openreview.net/forum?id=UERcQ…
Headed to #ICML2022 for the first in-person conference since pre-covid! Looking forward to exciting ML discussions with old & new friends Our workshop on Knowledge Retrieval and Language Models knowledge-retrieval-workshop.github.io is on Friday 22nd. Do stop by (or tune in online)! @icmlconf
we need more papers that don't report *any* results, instead focusing on refining shared vocabulary, e.g., what assumptions do different papers implicitly make about what LMs should be able to do? if you're in Punta Cana and interested, I'll be at Bohío Buffet at 7, let's chat!
🚨New pub🚨 How Climate Scenarios Lost Touch With Reality A failure of self-correction in science has compromised climate science’s ability to provide plausible views of our collective future by me & @jritch @ISSUESinST (online 7/26) Read it here now: drive.google.com/file/d/11UMwQ7…
New from Google Research: arxiv.org/abs/2106.16171 To build multilingual NLP systems, a successful recipe is to pre-train on a multilingual corpus, and then fine-tune on labeled data in a single transfer language -- usually English. But is English best?
New from Google Research! arxiv.org/abs/2102.01335 Show examples from the distribution you want, and our example extrapolator (Ex2) generates new examples from the same distribution. We use Ex2 for data augmentation, improving over SOTA methods on multiple NLP benchmarks! (1/3)
Two great, related papers to this: @adamjfisch, @kentonctlee, @mchang21, @JonClarkSeattle, Barzilay (2020). CapWAP. aclweb.org/anthology/2020… Zhao, Sharma, @TomerLevinboim, Soricut (2019). Informative Image Captioning with External Sources of Information. aclweb.org/anthology/P19-…
Real life example of the fact that words can label the function or purpose of items rather than the things themselves. (The relationship between images and associated language is diverse and highly goal dependent.)
This work was done together with a great set of collaborators: @kentonctlee , @mchang21 , @JonClarkSeattle , @BarzilayRegina. Code available online: github.com/google-researc….
Just read, and really enjoyed, @kentonctlee, @LuhengH, @ml_perception & @LukeZettlemoyer's coreference resolution and semantic role labeling papers from 2017 - if only all DL-for-NLP work was as clearly written and argued (with ablations) as these papers!
Excited to share new work w/ Ming-Wei Chang (@mchang21), Ice Pasupat (@IcePasupat) and Kristina Toutanova (@toutanova)! Compositional Generalization and Natural Language Variation: Can a Semantic Parsing Approach Handle Both? Paper link: ptshaw.com/nqgt5
Thrilled to share new work! “Retrieval-Augmented Generation for Knowledge-Intensive NLP tasks”. Big gains on Open-Domain QA, with new State-of-the-Art results on NaturalQuestions, CuratedTrec and WebQuestions. check out here: arxiv.org/abs/2005.11401. 1/N
Last summer we had a unique workshop/hackathon on technology for language documentation and revitalization. Here's a report of the various projects developed there, covering universal ASR, dictionary managers, corpus search tools, and social media bots: arxiv.org/pdf/2004.13203…
We will be holding a week-long development workshop on "Language Technology for Language Documentation and Revitalization" in Pittsburgh 8/12-16. Please consider joining if you're interested in building tools to help threatened/endangered languages sites.google.com/view/ltldr/home
Does "When did harry potter and sorcerer's stone movie come out?" look ambiguous? Ambiguity is inherent to open-domain QA. We introduce a new QA task for predicting question-answer pairs that represent different interpretations of the original question. arxiv.org/abs/2004.10645
This was an interesting paper to read. It's well-written, and the method that's used clearly works very well. A few things struck me as I read:
New from Google Research! REALM: realm.page.link/paper We pretrain an LM that sparsely attends over all of Wikipedia as extra context. We backprop through a latent retrieval step on 13M docs. Yields new SOTA results for open domain QA, breaking 40 on NaturalQuestions-Open!
@kentonctlee @IcePasupat @mchang21 Bonus result -- head-to-head comparison between two paradigms: implicit knowledge storage (e.g. BERT, GPT2, T5) vs explicit knowledge storage (e.g. REALM).
@kentonctlee @IcePasupat @mchang21 Instead of compressing massive textual knowledge into billions of uninterpretable parameters, REALM exposes the provenance for its representations and decision making. Corollary: we can control what REALM “knows” by hot-swapping the corpus.
Joint work with Kenton Lee (@kentonctlee), Zora Tung, Ice Pasupat (@IcePasupat) and Ming-Wei Chang (@mchang21)
New from Google Research! REALM: realm.page.link/paper We pretrain an LM that sparsely attends over all of Wikipedia as extra context. We backprop through a latent retrieval step on 13M docs. Yields new SOTA results for open domain QA, breaking 40 on NaturalQuestions-Open!