Daniel Israel @danielmisrael
PhD Student Studying AI/ML @UCLA Joined October 2011-
Tweets12
-
Followers240
-
Following2K
-
Likes76
After a successful presentation at the #ICML2023 main conference, delighted to share that ClimaX was also recognized with a Best Paper Award at the ICML Workshop on Scientific and ML modeling! Congrats to all coauthors @tungnd_13 @jo_brandstetter @akapoor_av8r @rejuvyesh!
After a successful presentation at the #ICML2023 main conference, delighted to share that ClimaX was also recognized with a Best Paper Award at the ICML Workshop on Scientific and ML modeling! Congrats to all coauthors @tungnd_13 @jo_brandstetter @akapoor_av8r @rejuvyesh! https://t.co/Kuep7xioN3
Modularity is critical for design of successful software and AI systems🧩. How can we apply this principle to foundation models for Reinforcement Learning? Our latest work, Decision Stacks, improves flexibility of RL through modular generative models. w/@adityagrover_ 🧵👇(1/8)
Introducing ClimaX, the first foundation model for weather and climate. A fast and accurate one-stop AI solution for a range of atmospheric science tasks. Paper: arxiv.org/abs/2301.10343 Blog: microsoft.com/en-us/research… Thread🧵 #ML #Climate #Weather #FoundationModel
📢Introducing ClimateLearn, a new PyTorch library for accessing climate datasets, state-of-the-art ML models, and high quality training and visualization pipelines. Blog: aditya-grover-group.github.io/blog/2023/clim… Docs: climatelearn.readthedocs.io Quickstart Colab: colab.research.google.com/drive/1WiNEK1B… 🧵 (1/n)
Excited to be co-organizing the first AI for Climate Science Bridge Program at #AAA23! Submit your papers by Nov 18. Further details here: ai4climatescience.github.io w/ @rejuvyesh @akapoor_av8r @manmeet3591 @niyogidev cc: @RealAAAI @ClimateChangeAI
Excited to be co-organizing the first AI for Climate Science Bridge Program at #AAA23! Submit your papers by Nov 18. Further details here: ai4climatescience.github.io w/ @rejuvyesh @akapoor_av8r @manmeet3591 @niyogidev cc: @RealAAAI @ClimateChangeAI https://t.co/8Zll9XjtEl
New paper out! We propose ConserWeightive BC, a simple but effective method for improving the performance and reliability of behavioral cloning methods such as DT and RvS in offline RL. @qqyuzu @adityagrover_
Samantha @craig_samantha6
259 Followers 3K FollowingSulin Liu @su_lin_liu
337 Followers 904 Following Postdoc @MIT. Machine Learning PhD @Princeton. Undergrad @NUSingapore.Kabir @kabirahuja004
448 Followers 415 Following CSE PhD Student @uwnlp | Ex-RF @MSFTResearch | cinephile 🎥Jinyuan (Tobias) @JinyuanWang7
78 Followers 767 Following Research Engineer @NUS. Data scientist @LushairZiqi Wang @wzq016
149 Followers 316 Following Ph.D. student @IllinoisCS, Prev undergrad @Tsinghua_Uni, Prev intern @GoogleTorti @Torti891912
0 Followers 163 Following My hobbies are reading, food and sports. I like cats😘 I like to meet new friends while traveling🎉🎉🎉Dante Blevins @DanteBlevi3226
101 Followers 3K FollowingCamilla Byrd @ByrdCamill7579
46 Followers 3K FollowingGail @taylorgail9356
402 Followers 3K FollowingMinhyuk Sung @MinhyukSung
583 Followers 466 Following Assistant professor @ KAIST https://t.co/sgDxHash1YCunxiang Wang @CunxiangWang
451 Followers 789 Following PhD Candidate @NLPwestlake, advised by Dr Yue Zhang. Interning @AWScloud. Research interest lies at Retrieval and LLMs. Seeking for a RS or PostDoc position.Xinyun Chen @xinyun_chen_
4K Followers 842 Following Research Scientist at @GoogleDeepMind. PhD from @Berkeley_EECS.Karush Suri @karush_
430 Followers 759 Following Meta Learner @Theteamatx in @GoogleAI swimmer & comic fan Past @borealisai @eceuoftAshima Suvarna 🌻 @suvarna_ashima
289 Followers 474 Following Phd-ing @UCLA | @Deepmind Scholar | Mitacs Scholar '19Yangqiu Song @yqsong
883 Followers 960 Following Associate Professor at HKUST, working on knowledge graphs, NLP, data mining on texts and graphsXubin Ren @xubinrencs
589 Followers 1K Following Ph.D. student of @hkudatascience and @HKUniversity Data Intelligence Lab, fortunately advised by @huang_chao4969. Trying to be a good data science researcher.Zhenrui Yue @Yueeeeeeee2837
715 Followers 992 Following Incoming Student Researcher @Google | PhD @UIUC | Prev @NVIDIA @ETH @UCSD @TUM | Interested in NLP, RecSys & IRHarsh Maheshwari @HarshMheshwari
1K Followers 1K Following Enthusiastic about #GenerativeAI #DataScience 🤖 | Constantly curious learner 🌱 | Applied scientist 2 at @amazon | Writer at @medium | @IITKGP GraduateJuba Ziani @JubaZiani
1K Followers 1K Following Assistant Prof @ Georgia Tech IsyE. Works on mechanism design, privacy, fairness, strategic ML. Postdoc @Penn, PhD @Caltech, he/himAaron Ewing @AEwing20613
9 Followers 678 FollowingLuke McDermott @lukemcdermotttt
112 Followers 452 Following AI Researcher at Modern Intelligence, focusing on Efficient Deep Learning & Adaptive AI | Incoming PhD @UCSanDiegoJan-Willem van de Mee.. @jwvdm
2K Followers 1K Following Associate Professor (UHD) at the University of Amsterdam; Probabilistic programming and its applications.Peiyang Song @p_song1
204 Followers 545 Following Honors CS Undergrad @UCSB_CCS. SURF fellow at Anima AI+Science Lab @caltech. Researcher @UCSBArchLab. Fmr @Tsinghua_Uni @NKU1919.Jacey Nixon @JaceyN88888
88 Followers 3K FollowingViacheslav Sinii @ummagumm_a
48 Followers 260 FollowingJirko Rubruck @RubruckJirko
83 Followers 149 Following PhD student @summerfieldlab and @saxelab. Interested in all learning from humans to machines.s @RmSuryansh
8 Followers 152 FollowingLinda @linda_cox56
327 Followers 3K FollowingMaximilian Du @du_maximilian
234 Followers 187 Following First-year Ph.D. student in @StanfordAILab, interested in causality, robustness, and self-directed play in AI, humans, & animals! (Oh, and I'm a writer too)Nandita Tomar @vyomaa_verse
109 Followers 450 Following brb NIT Bhopal CSE '24 | SDE Intern @FlipkartGunika Goyal @gunika_goyal
1 Followers 19 FollowingXinyue Wei @SarahWeii
148 Followers 130 Following PhD student at UCSD @HaoSuLabUCSD | Previous intern @AdobeResearchYanqiao ZHU @Zhu_Yanqiao
802 Followers 621 Following CS PhD at @UCLA, @NSF_CCAS. Organizer of @logconference. Graph & geometric representation learning / AI for ScienceYating Wu @YatingWu96
141 Followers 188 Following ECE Ph.D student @ UT Austin, advised by @jessyjli and @AlexGDimakis | テキサス大学に在籍する博士生Wei Liu @WeiLiu99
166 Followers 333 Following NLP ML Studying NLP now at @ShanghaiTechUni #NLProc | Incoming PhD @hkust | Prev. @AlibabaGroupNan HUO @NanHUO9637
208 Followers 420 Following CS PhD Student @HKUniversity. Previously M.S. @JohnsHopkins.Arif Ahmad @arif_ahmad_py
274 Followers 7K Following All things AI, Computer Science and Circuits! Prev. @GoogleAIZijian "Jason" Ding @jasonzding
1K Followers 1K Following PhD candidate @hcil_umd | Human Intent-AI Interaction Research | Incoming Intern @IBMResearch Previously @MSFTResearch @Dataminr @DesignLabUCSD @Cambridge_UniYin Fang @YinFang22900365
518 Followers 517 Following Ph.D. student in CS @ZJU_China. Looking for a post-doc position in AI4Science/LLM/KG. Feel free to reach me if you are interested in my research!Harsh Desai @dreamerharsh
1 Followers 3K FollowingMikaStars★ @MikaStars39_
170 Followers 629 Following Second year B.A. / B.S. in @ZJU_China Prev: Bsc in @Polytechnique Devoted in LLM Architecture & InterpretabilityHeming Xia @hemingkx
573 Followers 1K Following Ph.D. student @HongKongPolyU | Prev MEng & BSc @PKU1898 | Prev Intern @MSFTResearch (MSRA) | NLP | Language ModelingZeyun Lu 鲁泽沄 @zeyun_lu
270 Followers 404 Following Postdoc @KECKSchool_USC. Want to become a statsgen pro someday. Views my own.Omead Pooladzandi @omead_p
241 Followers 773 Following I enjoy second order optimization Machine Learning PhD @UCLA. ex Research Scientist Intern @AIatMeta (ideas are my own)Sébastien Darses @DarsesSebastien
3K Followers 3K Following Assoc. Prof. (he/him 🏳️🌈), Visiting @CRM_Montreal @CNRS_INSMI @CNRS 📚 Probability, Statistics, Analytic Number Theory 👨🎓Teaching project @highkholleWenhao Zhu @Wenhao_NLP
254 Followers 441 Following PhD candidate @NJUNLP, visiting PhD student @EdinburghNLP, interested in multilingual LLM and machine translation.Maggie Makar @Maggiemakar
4K Followers 628 Following Assistant prof @umichcse. Previously @MIT_CSAIL. Machine learning. Causal models. Healthcare. Swimming. @[email protected] Opinions are my own.Siyu Yuan @siyu_yuan_
228 Followers 258 Following Ph.D. candidate at Fudan University. Research Intern at @MSFTResearch Asia, Ex-Research Intern at @BytedanceTalk AI LabDaniel D'souza @mrdanieldsouza
535 Followers 904 Following Research Engineer @CohereForAI 💙 | @UMichECE Alum 〽️ | 🇮🇳✖️🇺🇸 💫"The Universe Works in Mysterious Ways"💫Heinrich Kuttler @HeinrichKuttler
2K Followers 698 Following Member of Founding Team @InflectionAI. Ex @FacebookAI, @DeepMind, @Google, @LMU_Muenchen, PhD math-ph. Opinions my own. (Can be yours for a small fee.)Siyuan Huang @siyuanhuang95
1K Followers 262 Following Research Scientist at BIGAI Working on #computer_vision and #3d_scene_understanding Ph.D. in Statistics from @UCLA Former intern at @DeepMind and @MetaAIMichi Yasunaga @michiyasunaga
3K Followers 867 Following CS PhD @Stanford working on language models and multimodal models. Previously @Meta @GoogleDeepMind @YaleMehran Kazemi @kazemi_sm
1K Followers 497 Following Senior Research Scientist @GoogleAI. Research areas: machine/deep learning, large language models, artificial general intelligence. Views my own.Weidi Xie @WeidiXie
2K Followers 577 Following Computer Vision Researcher. Associate Professor at SJTU, Previously @Oxford_VGG. 中文名:谢伟迪 Personal Webpage: https://t.co/sZoZ0AfKrXKimin @kimin_le2
1K Followers 330 Following Assistant professor at KAIST. Prev: Research scientist @GoogleAI, Postdoc @berkeley_ai & Ph.D at KAIST.Jiahui Lei @JiahuiLei1998
519 Followers 287 Following CS PhD student @Penn ; Previous Undergraduate at @ZJU_ChinaAlex Pan Vazquez @alexpan_neuro
454 Followers 1K Following Neuroscientist @PrincetonNeuro & @IntlBrainLab Reinforcement learning | Dopamine | Synaptic plasticity | Open scienceYotam Sagiv @SagivYotam
267 Followers 337 Following Grad student @PrincetonNeuro (@IlanaWitten/@NathanielDaw). Systems and computational neuro underlying RL. Also guitar, ballroom dance, cats, and naps. (He/him)Noémi Éltető @EltetoNoemi
618 Followers 431 Following AI-like human interested in human-like AI. Student researcher @DeepMind in London; PhD student @MPICybernetics in GermanyMiguel Saavedra @miguelSaaRuiz
350 Followers 363 Following PhD candidate at @UMontreal and @Mila_Quebec. Trying to blend robotics and AI.Jean-Christophe Gagno.. @GagnonAudet
287 Followers 400 Following Research engineer @Meta MTL | CS Msc @Mila_Quebec | Previously @IBMResearchAndrew Williams @CluelessAndrew
381 Followers 611 Following Kid with a bunch of questions, research @Mila_Quebec. MARL, QML, Edu x AI, interaction and intelligence, AI for common good.Arthur Ouaknine @ArthurOuaknine
341 Followers 432 Following Researcher fellow at @mcgillu and @Mila_Quebec working on forest monitoring. Core team member @ClimateChangeAI. Former PhD student @TelecomParis_ and @valeoaiRajaswa Patil @RajaswaPatil
2K Followers 3K Following Applied AI @getpostman || Prev. @Microsoft @ProseMsft @TCSResearch || Alumni @bitspilaniindia || Opinions are my ownShashank Srivastava @shsriva
480 Followers 5 Following Assistant Prof. at UNC Chapel Hill. I do research on language and AI. Previously @MSFTResearch, @mldcmuJunjie Hu @JunjieHu12
2K Followers 825 Following Assistant Professor @UWMadison PhD in #NLProc & #ML & #AI @LTIatCMU @SCSatCMU @CarnegieMellon. Interned @GoogleAI @MSFTResearch.UCSB NLP Group @ucsbNLP
1K Followers 729 Following The NLP Group @ University of California, Santa Barbara. Profs. @WilliamWangNLP, Xifeng Yan, Simon Todd, @CodeTerminator, @lileics; acct run by @m2saxonSebastian Ober @sebastian_ober
499 Followers 390 Following Senior Scientist for ML in Biologics Engineering at @AstraZeneca. Previously PhD student and @Gates_Cambridge scholar at @CambridgeMLG. Views are my own.Paula Torren Peraire @PTorrenPeraire
55 Followers 57 Following PhD student within AIDD consortium @AiddOneQingyun Wu @qingyun_wu
962 Followers 378 Following Assistant Professor at Penn State University @ISTatPENNSTATE Co-Creator of #AutoGen & #FLAMLSulin Liu @su_lin_liu
337 Followers 904 Following Postdoc @MIT. Machine Learning PhD @Princeton. Undergrad @NUSingapore.Aryan Deshwal @deshwal_aryan
201 Followers 602 Following PhD candidate in Computer Science at @WSUPullman. Previously research intern at @Meta and @Google.Viacheslav Borovitski.. @vabor112
135 Followers 88 Following ML Researcher @ ETH Zürich. PhD in Mathematics.Jędrzej Maczan @jedmaczan
215 Followers 163 Following https://t.co/TdsFvi4ymt 💙 @keep_FOMO_away. ASR research for dysarthric speech. Lead AI/ML @DNV_Group. Cover painting: https://t.co/rs0nMCNnN6Stephan Gouws @sgouws
1K Followers 1K Following AI Researcher, fascinated by communication across all modalities, wine lover, traveller, golfer. Prev: Google, DeepMind.Ulrich Paquet @ulrichpaquet
1K Followers 208 Following Research Scientist, Google DeepMind Director, African Institute of Mathematical Sciences (South Africa) Co-founder, Deep Learning IndabaAvishkar @avishkar58
2K Followers 592 Following Research Engineer at @Deepmind. Strengthen African machine learning through the Deep Learning Indaba @DeepIndaba. All opinions my own.Mark van der Wilk @markvanderwilk
4K Followers 592 Following Associate Professor in Machine Learning at the University of Oxford. Interested in automatic inductive bias selection using Bayesian tools.James Allingham @JamesAllingham
987 Followers 456 Following RS @GoogleDeepMind | Machine Learning PhD @CambridgeMLG | 🇿🇦Saurabh Kumar @drummatick
8K Followers 153 Following Building @kodomamo_JP | ML Engineer | Knows about 👉🏼 ML | AI | NLP | Flutter | Python Ex-Yahoo, Nokia, Rapyuta | @iitdelhi’20David Schlangen @davidschlangen
2K Followers 251 Following Professor, “Foundations of CompLing” @ https://t.co/L5gNujJTm3; & DFKI, Berlin. conversational (A)I, language & vision, interactive NLU, phil. of CLKabir @kabirahuja004
448 Followers 415 Following CSE PhD Student @uwnlp | Ex-RF @MSFTResearch | cinephile 🎥Jason Carman @jasonjoyride
21K Followers 2K Following Launching satellites @astranis & weekly startup documentaries via S³A meme page to check .. @memecrashes
353K Followers 252 Following currently working on our PhD (imaginary issues no one cares about except a really really really specific audience)Kangwook Lee @Kangwook_Lee
2K Followers 675 Following Assistant Professor, ECE, UW-Madison / Leading deep learning research @ KRAFTONBlake Bordelon ☕️.. @blake__bordelon
806 Followers 745 Following ML/Neuroscience PhD student at @HarvardTom Adamczewski @tmkadamcz
561 Followers 219 Following senior staff technology brother https://t.co/xyOnR8TIGk // https://t.co/6UX7kVY0Oe // https://t.co/alnSJHBGhzPeng Gao @penggaotweets
842 Followers 763 Following Assistant Professor at Virginia Tech CS; Postdoc at UC Berkeley; PhD at Princeton; Researching security, privacy, systems, and AIZiqi Wang @wzq016
149 Followers 316 Following Ph.D. student @IllinoisCS, Prev undergrad @Tsinghua_Uni, Prev intern @Google🏆Thrilled to share that VideoCon won the Best Paper Award at the Data Problems for Foundation Models #ICLR2024! I will present the work in 🇦🇹 Also, happy to share that I will be interning at @GoogleDeepMind w/ @kazemi_sm this summer! Happy to connect with folks in ICLR.
📢 📽✍️We introduce VideoCon, a video-text dataset for training SOTA alignment model. It resolves a typical issue in video-text alignment models that struggles with robustness. w/ @YonatanBitton, Idan Szpektor, @kaiwei_chang , @adityagrover_ video-con.github.io 🧵 1/
@danielmisrael @yuzhaouoe Yeah! @yuzhaouoe and I were chatting about it earlier today, it's really clever!
Llama 3 was trained using intra-document causal masking, as suggested by @yuzhaouoe's paper "Analysing The Impact of Sequence Composition on Language Model Pre-Training"! 🚀🚀🚀 arxiv.org/abs/2402.13991
@PMinervini @yuzhaouoe you may also find our recent work interesting, in which we using "packing" to dramatically speed up inference x.com/siyan_zhao/sta… arxiv.org/pdf/2404.09529……
🚨LLM RESEARCHERS🚨Want a free boost in speed and memory efficiency for your HuggingFace🤗LLM with ZERO degradation in generation quality? Introducing Prepacking, a simple method to obtain up to 6x speedup and 16x memory efficiency gains in prefilling prompts of varying lengths.…
How to identify bias in language agency?Eg. in texts describing White men as “leading” & Black women as “helping”?🧐 🔎String matching?❌NO! 🔎Sentiment classifier?❌No! ✅Our agency classifier CAN! It reveals gender, racial, and intersectional bias🤯 🔗: arxiv.org/abs/2404.10508
一种名为 Prepacking 的简单方法,用于加速 LLM 在推理过程中的预填充和增加吞吐量 作者来自 UCLA: Siyan Zhao∗, Daniel Israel∗,Guy Van den Broeck, Aditya Grover 核心点: 1. 在推理过程中,Prepacking…
🚨LLM RESEARCHERS🚨Want a free boost in speed and memory efficiency for your HuggingFace🤗LLM with ZERO degradation in generation quality? Introducing Prepacking, a simple method to obtain up to 6x speedup and 16x memory efficiency gains in prefilling prompts of varying lengths.…
@danielmisrael seems like this would be a huge benefit for everyone concerned with performance and quality
Very excited about this work! If you are an LLM researcher frustrated by long wait times on generations, I highly recommend you to check out prepacking.
🚨LLM RESEARCHERS🚨Want a free boost in speed and memory efficiency for your HuggingFace🤗LLM with ZERO degradation in generation quality? Introducing Prepacking, a simple method to obtain up to 6x speedup and 16x memory efficiency gains in prefilling prompts of varying lengths.…
(11/n) We're making our code publicly available on Github. Dive into our implementation and see how prepacking can make your LLMs faster and leaner. Github: github.com/siyan-zhao/pre… Paper: arxiv.org/abs/2404.09529 Awesome collaborations with co-first author @danielmisrael under…
(10/n) Beyond prefilling, the concept of packing holds great promise for LLM generation. Our preliminary results on a toy batch show prepacking can dramatically reduce memory waste by bin-packing KV caches.
(9/n) Moreover, prepacking is fully implemented in PyTorch and is hardware agnostic. This means you can easily integrate it into your existing workflow, bringing the benefits of efficient LLM inference to a wide range of applications without the need to write custom CUDA kernels.
(8/n) We can estimate speedup from two statistics: 1) batch size reduction after prepacking and 2) max absolute deviation, the maximum length of a batch deviates from the mean length. These metrics can predict the speedup obtained by using prepacking over full batching.
(7/n) Thus far we have discussed packing over batches, but we also explore “Dataset Prepacking”, which is applying prepacking on the whole dataset. We show Dataset Prepacking outperforms the baselines in prefilling times.
(6/n) Prepacking reduces computation on padded tokens, significantly reducing peak GPU memory usage, allowing batch sizes up to 16x larger without encountering out-of-memory errors. This efficiency is crucial for maximizing hardware utilization in resource-constrained settings.
(5/n) Prepacking exhibits greater speedup gains on larger batches of increased prompt length diversity, causing more padding overhead for Full Batching. Across batch sizes for quantized Llama2 models, prepacking achieves substantial speedup up to 6x over Full Batching on an A6000
(4/n) We apply prepacking in a variety of settings across 6 diverse language datasets and 4 language models, and we inevitably see the same thing each time: a dramatic speed up in prefilling and time-to-first-token (TTFT) compared to Full Batching (standard HF implementation).
(3/n) The issue with packing prompts together is they will attend to each other. We introduce independent masking and restart positional encoding. These techniques make a single forward pass on multiple prompts in sequence equivalent to a forward pass on each prompt separately.
(2/n) Prefilling in transformer-based LLMs wastes computational resources on pad tokens for variable-length prompts. As LLMs increasingly support up to 10 million tokens, it is a growing inefficiency. Instead, Prepacking combines prompts of varying lengths into compact batches.
🚨LLM RESEARCHERS🚨Want a free boost in speed and memory efficiency for your HuggingFace🤗LLM with ZERO degradation in generation quality? Introducing Prepacking, a simple method to obtain up to 6x speedup and 16x memory efficiency gains in prefilling prompts of varying lengths.…
Good example of how damaging the bar is for the viewer. Position looks amazing for Black, up a pawn, active queen and yet the bar is all the way up. Why?