Eric Chu @its_ericchu
Research scientist @ Google DeepMind. AI reasoning + alignment/safety to help humans. Gemini, Bard, PaLM 2. Prev PhD @ MIT. web.media.mit.edu/~echu 🇺🇸🇬🇧 Joined May 2021-
Tweets223
-
Followers2K
-
Following794
-
Likes3K
Gemini 1.5 Pro - A highly capable multimodal model with a 10M token context length Today we are releasing the first demonstrations of the capabilities of the Gemini 1.5 series, with the Gemini 1.5 Pro model. One of the key differentiators of this model is its incredibly long…
Seeing some confusion like: "You trained a model to do Bad Thing, why are you surprised it does Bad Thing?" The point is not that we can train models to do Bad Thing. It's that if this happens, by accident or on purpose, we don't know how to stop a model from doing Bad Thing 1/5
Seeing some confusion like: "You trained a model to do Bad Thing, why are you surprised it does Bad Thing?" The point is not that we can train models to do Bad Thing. It's that if this happens, by accident or on purpose, we don't know how to stop a model from doing Bad Thing 1/5
Thrilled to share results from our 5-month field experiment on the effects of generative AI on entrepreneurial performance among Kenyan entrepreneurs: osf.io/preprints/osf/… joint with @RowanPClarke, @solenedelecourt, @daveholtz, @orgRem. 1/7
Super excited about this new paper from our group: using LMs to help "auto-formalize" sequential decision-making problems via low-level interaction with an environment. (something weird happened with a link shortener in the first tweet; paper is arxiv.org/abs/2312.08566)
Super excited about this new paper from our group: using LMs to help "auto-formalize" sequential decision-making problems via low-level interaction with an environment. (something weird happened with a link shortener in the first tweet; paper is arxiv.org/abs/2312.08566)
Update: I recently joined @liyuajia and @OriolVinyalsML’s team in @GoogleDeepMind. Been a fun year at Alphabet, and lots more to do to make LLMs useful & deployable! 🙋🏻♂️🧠⚒️🚢 I’m at the Alignment workshop/#NeurIPS2023, let’s chat about alignment🤝capabilities work + be friends
Update: I recently joined @liyuajia and @OriolVinyalsML’s team in @GoogleDeepMind. Been a fun year at Alphabet, and lots more to do to make LLMs useful & deployable! 🙋🏻♂️🧠⚒️🚢 I’m at the Alignment workshop/#NeurIPS2023, let’s chat about alignment🤝capabilities work + be friends
the gemini team moves very fast and the energy is contagious! this is going to break a lot of narratives and assumptions in the coming year :) s/o to @LencKarel @antoine77340 and @drjwrae for pairing on this!
the gemini team moves very fast and the energy is contagious! this is going to break a lot of narratives and assumptions in the coming year :) s/o to @LencKarel @antoine77340 and @drjwrae for pairing on this!
Gemini's out! ♊️ Also Bard starts using Gemini Pro today (with Gemini Ultra coming early next year)
Gemini's out! ♊️ Also Bard starts using Gemini Pro today (with Gemini Ultra coming early next year)
Many good reasons to hillclimb on reasoning/STEM datasets w closed form solutions. Still, a worthy direction is reasoning in openended domains (see paper)! Imo the Q to ask is why self-consistency would (or wouldn't) work. More creative, non-majority voting mechanisms may help…
Many good reasons to hillclimb on reasoning/STEM datasets w closed form solutions. Still, a worthy direction is reasoning in openended domains (see paper)! Imo the Q to ask is why self-consistency would (or wouldn't) work. More creative, non-majority voting mechanisms may help…
Creative! "Visual anagrams" / optical illusions. Nice example of work (with appealing end results) that requires an understanding of the underlying method (in this case diffusion models). Reminds me of the OG neural style transfer paper by Gatys et al in that way.
Creative! "Visual anagrams" / optical illusions. Nice example of work (with appealing end results) that requires an understanding of the underlying method (in this case diffusion models). Reminds me of the OG neural style transfer paper by Gatys et al in that way.
Introducing Pika 1.0 ! I’m thrilled to share this video that demonstrates how powerful our new product is. Create and edit amazing videos simply by typing. Sign up at pika.art
Announcing the L3 Lab at CMU! cmu-l3.github.io We focus on Learning, Language, and Logic, including: - Principles of ML for language - ML in high-trust areas, such as verifying math and programs - ML systems that improve over time Recruiting PhD students for fall 2024!
🚀Alignment improved! 🌐 Our RL fine-tuning group at @GoogleDeepMind presents a new algorithm for LLM fine-tuning—say hello to IPO: Identity Preference Optimization! 🌟
🚀Alignment improved! 🌐 Our RL fine-tuning group at @GoogleDeepMind presents a new algorithm for LLM fine-tuning—say hello to IPO: Identity Preference Optimization! 🌟
Go work with Martin!! great guy and mentor, curious about everything, and I learned a lot from him. also has a cute dog Uses deep knowledge of ML/NLP/causal inference/networks to study social platforms. And for the ppl here excited about Community Notes, check out Martin's work
Go work with Martin!! great guy and mentor, curious about everything, and I learned a lot from him. also has a cute dog Uses deep knowledge of ML/NLP/causal inference/networks to study social platforms. And for the ppl here excited about Community Notes, check out Martin's work
Nice thread giving added perspective on google deepmind’s music generation tools, coming from a long time AI + music and audio expert @keunwoochoi deepmind.google/discover/blog/…
Nice thread giving added perspective on google deepmind’s music generation tools, coming from a long time AI + music and audio expert @keunwoochoi deepmind.google/discover/blog/…
David Attenborough is now narrating my life Here's a GPT-4-vision + @elevenlabsio python script so you can star in your own Planet Earth:
“J. COLE: I actually started off majoring in computer science, but I knew right away I wasn't going to stay with it.. I had this one professor who was the loneliest, saddest man I've ever known. He was a programmer, and I knew I didn't want to do whatever he did” 🥲 Cole world
Cool work by @eghbal_hosseini and team on how language models predict by linear extrapolation in representation space. Should be of interest to the mechanistic interpretability and possibly hallucination/factuality communities!
Cool work by @eghbal_hosseini and team on how language models predict by linear extrapolation in representation space. Should be of interest to the mechanistic interpretability and possibly hallucination/factuality communities!
Chance to join a new AI safety lab at Scale - I worked closely with Summer on RLHF at Google, and she’s great to work with! Topics include robust evals, automated redteaming, scalable oversight. Could imagine many more, especially given its spot in the data ecosystem
Chance to join a new AI safety lab at Scale - I worked closely with Summer on RLHF at Google, and she’s great to work with! Topics include robust evals, automated redteaming, scalable oversight. Could imagine many more, especially given its spot in the data ecosystem
Luca Soldaini 🎀 @soldni
6K Followers 1K Following I like tokens! Lead for OLMo data team at @allen_ai (Dolma 🍇), OSS is fun, @QueerInAI organizer 🤖☕️🍕they/them (views mine, not my employer’s)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/2A3qF5vdJwKayo Yin @kayo_yin
8K Followers 560 Following PhD student @berkeley_ai @berkeleynlp working on interpretability and signed languages. Former @msftresearch @deepmind @carnegiemellon @polytechnique. 🇫🇷🇯🇵Kyunghyun 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).Sara Hooker @sarahookr
39K Followers 7K Following I lead @CohereForAI. Formerly Research @Google Brain @GoogleDeepmind. ML Efficiency at scale, LLMs, @trustworthy_ml. Changing spaces where breakthroughs happen.Talia Ringer 🟣 �.. @TaliaRinger
26K Followers 6K Following Professor, @plfmse, @IllinoisCS! Proof Automation. @SigplanM & CCF Founder. Israeli-American for peace, equality, & justice. They/היא, ND, bi. די לכיבושDanish Pruthi @danish037
7K Followers 628 Following Faculty at Indian Institute of Science, Bangalore. PhD from @LTIatCMU.Sasha Luccioni, PhD �.. @SashaMTL
19K Followers 4K Following AI & Climate @HuggingFace, Board Member of @WiMLworkshop and @ClimateChangeAI. @techreview 35 Innovators under 35, @TEDTalks speaker. She/her/Dr/ 🦋Miles Brundage @Miles_Brundage
43K Followers 10K Following Policy research at @openai. I mostly tweet about AI, animals, and sci-fi. He/him. Views my own.Niloofar (Fatemeh) Mi.. @niloofar_mire
4K Followers 1K Following Postdoc @uwcse-@uwnlp, Ph.D. from @ucsd_cse /Privacy, ML, NLP, @winlpworkshop chair, @MSFTResearch - Semantic MachinesEkin Akyürek @akyurekekin
2K Followers 727 Following graduate student in computer science @MITEECS/@MIT_CSAIL👩💻 Paige Bai.. @DynamicWebPaige
59K Followers 2K Following ✨Keep it simple, make it scale. AI should be about empowering people, building understanding, & making dreams realities. 👩💻GenAI @GoogleDeepMind ex-@GitHubAakanksha Chowdhery @achowdhery
7K Followers 3K Following LLMs @ Google DeepMind :: PaLM, Gemini // Previously @MSFTResearch, @Stanford, @Princeton // views my own and subject to changeEugene Vinitsky @EugeneVinitsky
13K Followers 2K Following Lets make multi-agent learning easy. Anti-cynic. RS at Apple, Asst. Prof at @nyutandon. He/him.Laura Ruis @LauraRuis
3K Followers 638 Following Currently research intern @cohere, PhD supervised by @_rockt and @egrefen. Language and LLMs. Spent time at FAIR, Google, and NYU (@LakeBrenden). She/her.Divy Thakkar @divy93t
5K Followers 2K Following Strategy, Programs & Product @GoogleAI , HCI Researcher. Ph.D @CityUniLondon Alumni @iift1963 @daiictofficial. Personal views.rohan anil @_arohan_
12K Followers 2K Following Principal Engineer, @GoogleDeepMind Gemini. prev PaLM-2. Tinkering with optimization and distributed systems. opinions are my own.Nathan Benaich @nathanbenaich
51K Followers 32K Following solo member of investment staff @airstreet, brewing ambition @airstreetcafe, next token predictor @airstreetpressmerve @mervenoyann
56K Followers 4K Following open-sourceress at @huggingface 🧙🏻♀️ proud mediterrenean 🍋 I do TL;DR on ML papersSaurabh Srivastava @_saurabh
832 Followers 376 Following Research in reasoning for better program synthesis (PhD, Postdoc, YC)Kasha Kretz @KashaKretz86414
64 Followers 5K FollowingJindong Gu @Jindong73504766
296 Followers 891 Following Senior Researcher @UniofOxford, Faculty Researcher @GoogleResearch, PhD @LMU_Muenchen #ResponsibleAI #AISafety #GenAI Homepage: https://t.co/YOSVO3jb6hArianna Manzini @Arianna_Manzini
1K Followers 935 Following Ethics Research Scientist @DeepMind | Prev: Postdoc in #ethics #robots @BristolUni and PhD in #ethics #genomics @UniofOxfordThalia Sherbo @SherboThal19550
93 Followers 5K FollowingAbdulrahman Tabaza @embed_dim
4 Followers 809 Following enjoyer of various vector spaces, encoders and modalitiesDemetria Kraskouskas @KraskouskDemetr
108 Followers 5K FollowingVi @AvimanyuRoy3
579 Followers 2K Following 🍎🕊/🦦☕️/😴🛌/he/him Shouting into the Void (TM) GPU poor peasantPenelope Neumeister @PenelopeNe71084
86 Followers 5K FollowingLia Heatley @HeatleyLia43940
87 Followers 5K FollowingZhiqing Sun @EdwardSun0909
2K Followers 1K Following CS PhD @LTIatCMU working on scalable alignment. BS @PKU1898Ivy-rose Flathers @IvyFlather97955
54 Followers 5K FollowingSohail khan @mrshyspy
89 Followers 2K Following MERN Stack Developer | Documenting my coding journey | Immediate job seeker | Let's connect!Ricki Salloum @RicSallo
81 Followers 5K FollowingNasle @Nasle471810
125 Followers 2K FollowingGreg Wilson @wilsonthegreg
203 Followers 432 Following The Growth Guy | Tweets on business and productivity.Carina Callihan @CariCalli
19 Followers 3K FollowingShubhashis Roy Dipta @iamdipta007
248 Followers 1K Following PhD @umbc || Multimodal (NLP + CV) || 🏠 https://t.co/XFDVDULgwS || 📝 https://t.co/UaVN46IIe4 || ✍️ https://t.co/PxOoQefIDdAnita @Anita4896519423
33 Followers 997 FollowingEvia Spiegle @ev_spie
14 Followers 3K FollowingGuadalupe Dearmitt @GuadalupeD32744
80 Followers 5K FollowingKeavie Morr @keav_m
27 Followers 5K FollowingDaniel Han @danielhanchen
7K Followers 941 Following Building @UnslothAI. Finetune LLMs 30x faster https://t.co/aRyAAgKOR7. Prev ML at NVIDIA. Hyperlearn used by NASA. I like maths, making code go fastTiana Igo @tia_igo
103 Followers 5K FollowingZoe @zoe_wong00
6 Followers 62 FollowingJoe @joemkwon
760 Followers 2K Following thinking about what good futures (embedded with powerful AI systems) might look likeSatyam @SatyamGuptaDev
454 Followers 5K Following Building Websites and solving LeetCode | React.js • Next.js • AI | Let's build great productsNnamaka Dike @DNnamaka
104 Followers 359 Following Machine Learning Engineer | Software Engineering. Graduate @alx_africa #mlengineer #machinelearningMaya Deniken @deniken15451
45 Followers 5K FollowingJunko Neizer @JunkoNeize94719
52 Followers 5K FollowingSreejith Krishnan R @skr_research
0 Followers 332 FollowingTyler Lewis @TylerLewis8TM
16 Followers 685 Following Active trade analyst with 8 traders markets, Top tier portfolio manager, Copy trade initiative and Hedge Funds Advisory Group Committee.Yacine Jernite @YJernite
4K Followers 1K Following ML & Society lead @huggingface, NLPer at heart, focusing on data and ML systems governance these days he/him #BlackLivesMatterSakuneson @sakuneson32425
57 Followers 2K FollowingAntoine Yang @AntoineYang2
707 Followers 411 Following Research Scientist @GoogleDeepMind, Gemini multi-modal 💎. Prev: PhD @Inria & @ENS_ULM, MEng @Polytechnique.Feryal @FeryalMP
9K Followers 2K Following Staff Research Scientist @DeepMind & Board of Directors @WiMLworkshop.AK @_akhaliq
310K Followers 3K Following AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo follow on Hugging Face: https://t.co/q2Qoey80GxJim Fan @DrJimFan
229K Followers 3K Following @NVIDIA Sr. Research Manager & Lead of Embodied AI (GEAR Lab). Creating foundation models for Humanoid Robots & Gaming. @Stanford Ph.D. @OpenAI's first intern.Luca Soldaini 🎀 @soldni
6K Followers 1K Following I like tokens! Lead for OLMo data team at @allen_ai (Dolma 🍇), OSS is fun, @QueerInAI organizer 🤖☕️🍕they/them (views mine, not my employer’s)Google DeepMind @GoogleDeepMind
944K Followers 275 Following We’re a team of scientists, engineers, ethicists and more, committed to solving intelligence, to advance science and benefit humanity.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/2A3qF5vdJwSam 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.Natasha Jaques @natashajaques
25K Followers 1K Following Senior Research Scientist at @GoogleAI and Assistant Professor @uwcse. Social Reinforcement Learning in multi-agent and human-AI interactions. PhD from @MIT.Kayo Yin @kayo_yin
8K Followers 560 Following PhD student @berkeley_ai @berkeleynlp working on interpretability and signed languages. Former @msftresearch @deepmind @carnegiemellon @polytechnique. 🇫🇷🇯🇵Kyunghyun 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).Thomas Wolf @Thom_Wolf
68K Followers 4K Following Co-founder and CSO @HuggingFace - open-source and open-scienceAnthropic @AnthropicAI
262K Followers 26 Following We're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant Claude at https://t.co/aRbQ97uk4d.Jeff 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)Sara Hooker @sarahookr
39K Followers 7K Following I lead @CohereForAI. Formerly Research @Google Brain @GoogleDeepmind. ML Efficiency at scale, LLMs, @trustworthy_ml. Changing spaces where breakthroughs happen.Yi Tay @YiTayML
29K Followers 97 Following chief scientist / cofounder @RekaAILabs 🫠 past: research scientist @google brain 🤯 currently learning to be a dad 🍼Talia Ringer 🟣 �.. @TaliaRinger
26K Followers 6K Following Professor, @plfmse, @IllinoisCS! Proof Automation. @SigplanM & CCF Founder. Israeli-American for peace, equality, & justice. They/היא, ND, bi. די לכיבושDanish Pruthi @danish037
7K Followers 628 Following Faculty at Indian Institute of Science, Bangalore. PhD from @LTIatCMU.Sasha Luccioni, PhD �.. @SashaMTL
19K Followers 4K Following AI & Climate @HuggingFace, Board Member of @WiMLworkshop and @ClimateChangeAI. @techreview 35 Innovators under 35, @TEDTalks speaker. She/her/Dr/ 🦋Sarah Wang @sarahdingwang
9K Followers 904 Following General Partner @a16z growth fund. “Excellence is the capacity to take pain”Dion Almaer @dalmaer
28K Followers 6K Following Bringing nuance to a knife fight, surprisingly older than @bgalbs. Product @augmentcode. Prev: Google, Shopify, Ajaxian & more. 🌥: @almaer.com 🧵: dionalmaerAugment @augmentcode
650 Followers 8 Following Augment's expert understanding of your codebase and dependencies removes the toil in your day, so you experience the joy of coding.Ed Zitron @edzitron
78K Followers 5K Following Newsletter https://t.co/D5qDgUKaNR - Better Offline Podcast - https://t.co/pUoGsuaQTw - Columnist Business Insider - CEO at https://t.co/5idt8AyPqr - Award-Winning Tech PRArianna Manzini @Arianna_Manzini
1K Followers 935 Following Ethics Research Scientist @DeepMind | Prev: Postdoc in #ethics #robots @BristolUni and PhD in #ethics #genomics @UniofOxfordCanfer Akbulut @canfer_akbulut
173 Followers 131 Following sociotechnical AI research @googledeepmindKristian Lum @KLdivergence
22K Followers 1K Following Research Scientist at Google DeepMind | @FAccTConference OG | Past Twitter META, @hrdag & UPenn, UChicago faculty |Ed Newton-Rex @ednewtonrex
7K Followers 1K Following CEO of @fairlytrained / Composer. Previously founded Jukedeck, VP Audio at Stability AI.Aleksandra Faust @AleksandraFaust
2K Followers 515 Following Research Scientist with Google @Deepmind. Previously, @GoogleAI in #GoogleBrain. @Waymo, @SandiaLabs, @UNM, @UIUC.Tejas Kulkarni @tejasdkulkarni
19K Followers 1K Following CEO @CSM_ai. Amplifying 3D creativity. Discord: https://t.co/OoSpHkaelA. Former: Scientist @GoogleDeepMind. PhD @MITArjun Panickssery is .. @panickssery
1K Followers 2K Following Researching scalable oversight @MATSprogram | prev @METR_Evals @ai_risks | spaced repetition | AI safety | https://t.co/p887k6EsFsVincent Conitzer @conitzer
4K Followers 1K Following AI professor. Director, @FOCAL_lab @CarnegieMellon. Head of Technical AI Engagement, @UniofOxford @EthicsInAI. Author, "Moral AI - And How We Get There."Mayank Agrawal @_magrawal
501 Followers 255 Following building @roundtabledotai prev: comp cog neuro phd @ princeton, xc + t&f @ swarthmoreWill Merrill @lambdaviking
2K Followers 570 Following Ph.D. student @ NYU🗽 Theoretical aspects of NLP and LMs /nætʃɹəl/🇮🇸 + formal🤵 languages + TCS🧮Chris Paxton @chris_j_paxton
8K Followers 1K Following Mostly posting about robots. Embodied AI @hellorobotinc, formerly @AIatMeta, @NVIDIAAI, @zoox. All views my own.Del Johnson @DelJohnsonVC
23K Followers 4K Following "The Most contrarian thinker in VC" & Father of Modern Venture Capital. VC, Angel, LP. prev: @Google @Oracle @ucberkeley @ColumbialawZhiqing Sun @EdwardSun0909
2K Followers 1K Following CS PhD @LTIatCMU working on scalable alignment. BS @PKU1898fforres @fforres
5K Followers 5K Following Human Data @OpenAI - Doin @JSconfCL & @JavaScriptChile. Lovin' Frontend Infrastructure, JS, DX, TS — @_pilliin_'s husband — he/him — Living with ADHD ❤️Crémieux @cremieuxrecueil
88K Followers 907 Following I write about genetics, 'metrics, and demographics. Read my long-form writing at https://t.co/8hgA4nNS2A.Alan Chan @_achan96_
857 Followers 1K Following PhD student @Mila_quebec || Research Scholar @GovAI_ || AI safety || 🇨🇦Jon Chu @heyjchu
2K Followers 396 Following Partner @khoslaventures, founder @ Koality (exited), OG @ PLTR, OPEN, Docker, ML @ FBShunyu Yao @ShunyuYao12
7K Followers 858 Following Language agents (ReAct, Reflexion, Tree of Thoughts) for digital automation (WebShop, SWE-bench, SWE-agent)noahdgoodman @noahdgoodman
2K Followers 109 Following Professor of natural and artificial intelligence @Stanford. Research Scientist at @GoogleDeepMind. (@StanfordNLP @StanfordAILab etc)Desh Raj @rdesh26
3K Followers 2K Following Research Scientist @Meta (AI Speech) | Previously: @jhuclsp, @IITGuwahatiAlexandr Wang @alexandr_wang
143K Followers 697 Following ceo at @scale_ai. rational in the fullness of timeKawin Ethayarajh @ethayarajh
3K Followers 730 Following PhD student @StanfordAILab @stanfordnlp Working on machine learning under human incentives.Meaning Alignment Ins.. @meaningaligned
936 Followers 10 Following The Meaning Alignment Institute is a research organization with the goal of ensuring human flourishing in the age of AGI.Corry Wang @corry_wang
25K Followers 254 Following Strategy @ Google | Formerly tech equity research @ Bernstein Research. All opinions expressed are my own, and do not represent Google'sderek guy @dieworkwear
846K Followers 963 Following Menswear writer. Editor at @putthison. Creator of @RLGoesHard. Bylines at The New York Times, The Washington Post, The Financial Times, Esquire, and Mr. PorterSimon Schug @ssmonsays
450 Followers 307 Following PhD student working on meta-learning et al. @[email protected]Hume @hume_ai
16K Followers 17 Following Empathic AI research lab. Building AI with emotional intelligence: https://t.co/BuyjmutoBhSamuel Hammond 🌐�.. @hamandcheese
22K Followers 2K Following Senior economist @joinFAI. Nonresident fellow @NiskanenCenter. Pluralist. 'The world is second best, at best.' | [email protected]Jason Baldridge @jasonbaldridge
10K Followers 1K Following Research scientist at Google in Austin working on grounded language understanding. [email protected]swyx @swyx
91K Followers 3K Following Anti-ego ideas for anti-ergodic life. Founder, @smolmodels ▹ Listen: @latentspacepod ▹ Read: @coding_career ▹ Join: @aiDotengineerLatent Space Podcast @latentspacepod
8K Followers 43 Following The first place over 50k AI Engineers gather to talk models, tools and ideas. Breaking news today you will use at work tomorrow! Hosted by @swyx and @fanahovaYisong Yue @yisongyue
19K Followers 3K Following Machine Learning @Caltech (@YueLabCaltech). AI for invention at @AsariAILabs. Autonomous Driving at https://t.co/riZHAmvcAr. Senior Program Chair @iclr_conf.Xuhui Zhou @nlpxuhui
689 Followers 431 Following PhD student @LTIatCMU. Previously, @GeorgiaTech, @UWNLP, and @Apple. Social Intelligence in language +X. He/Him.🐳Saurabh Srivastava @_saurabh
832 Followers 376 Following Research in reasoning for better program synthesis (PhD, Postdoc, YC)Sakana AI @SakanaAILabs
19K Followers 0 Following We are a Tokyo-based R&D company on a quest to create a new kind of foundational AI model based on nature-inspired intelligence. https://t.co/LonvHEtlJRDaniel Han @danielhanchen
7K Followers 941 Following Building @UnslothAI. Finetune LLMs 30x faster https://t.co/aRyAAgKOR7. Prev ML at NVIDIA. Hyperlearn used by NASA. I like maths, making code go fastHan @hhua_
3K Followers 4K Following Invest @GVteam during 🌞 and hacker at 🌒. Investing in AI, infra, deep tech, fintech/crypto ⚡️🤖🧠. Views are my own.Pere Rosselló @PeRossello
6K Followers 305 Following Astrophysics student at Universidad de La Laguna, Tenerife (Spain). All media posted here is my own (unless otherwise stated).DoNotPay @donotpay
70K Followers 346 Following Your A.I. consumer champion. Helping millions of consumers resolve their problems.Cognition @cognition_labs
123K Followers 19 Following Makers of Devin, the first AI software engineer. We are an applied AI lab focused on reasoning, and code is just the beginning. Join us: https://t.co/tpfZwEwGiqNever thought I’d leave Google but an exciting new challenge came up and I knew I had to take it. Almost 6 years. 4 teams - Search, Research, X - the moonshot factory, Google Labs, and finally Workspace. Thank you again to the amazing people that I’ve worked with. ✌🏼
Solution: Gated SAEs have two encoders, one to find which features are active, the other to estimate active features' magnitudes. The L1 penalty only applies to the first. This still works if you tie most of the weights of the two encoders, making this cheap to run.
It's not PPO > DPO, It's policy generated data > stale data, In this paper, we answer this question by performing a rigorous analysis of a number of fine-tuning techniques on didactic and full-scale LLM problems. Our main finding is that, in general, approaches that use…
What a year it has been at @augmentcode! Today we have reached a massive milestone on our journey to augment software engineers with AI: We've secured $252M in Series B funding! I am proud to be part of the team and excited about what the future holds. techcrunch.com/2024/04/24/eri…
Are LLMs biased toward themselves? Frontier LLMs give higher scores to their own outputs in self-eval. We find evidence that this bias is caused by LLM's ability to recognize their own outputs This could interfere with safety techniques like reward modeling & constitutional AI
✨Excited to finally drop our new paper: SSMs “look like” RNNs, but we show their statefulness is an illusion🪄🐇 Current SSMs cannot express basic state tracking, but a minimal change fixes this! 👀 w/ @jowenpetty, @Ashish_S_AI arxiv.org/abs/2404.08819
🔥Thrilled to introduce HypoGeniC: Hypothesis Generation with Large Language Models 🔥 How can LLMs systematically propose and verify hypotheses based on observations for #ScientificDiscovery? Read our paper to find out! 📄: arxiv.org/abs/2404.04326… Details in 🧵 (1/n):
Some statistics on the superalignment fast grants: We funded 50 out of ~2,700 applications, awarding a total of $9,895,000. Median grant size: $150k Average grant size: $198k Smallest grant size: $50k Largest grant size: $500k Grantees: Universities: $5.7m (22) Graduate…
🌟Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision 🌟 arxiv.org/abs/2403.09472 How can we keep improving AI systems when their capabilities surpass those of human supervisors? (1/n)
finished listening to this audio book today and just got the physical copy. what a book. highly recommend for all cs/hci folks👀
🎮 Introducing the new and improved Policy-Guided Diffusion! Vastly more accurate trajectory generation than autoregressive models, with strong gains in offline RL performance! Plus a ton of new theory and results since our NeurIPS workshop paper... Check it out ⤵️
Language models struggle to search, not due to an architecture problem, but a data one! They rarely see how to search or backtrack. We show how LLMs can be taught to search by representing the process of search in language as a flattened string, a stream of search (SoS)!
I’ve been writing a lot of sci-fi set in posthuman worlds lately, and thinking about new types of conflict that might open up. Some ideas in comic form (with explanations and examples below):
New paper finds GPT-4 is very good at pricing, acting as an agent to get a merchant the best price. In fact, it is a little too good. When it is possible to establish an oligarchy, LLM agents spontaneously collude on pricing to the detriment of customers! arxiv.org/abs/2404.00806
Auto-discovering circuits for auto-discovered LLM behaviors! feature-circuits.xyz
But LMs have many difficult-to-anticipate behaviors and mechanisms. Can we get circuits for those as well? Yes! By combining our method with @ericjmichuad_/@tegmark’s quanta discovery methods, we automatically discover thousands of feature circuits from raw text data.
Interesting, this is actually more or less what we did for StarCraft as well. It was crucial for us to condition the model on the MMR of the opponent that were playing against both during initial imitation learning and also the RL fine-tuning stage.
A good next token predictor would predict legal but low skill moves if the game begins with random moves. This is what we find. But, we can derive a skill vector and add it to the model, increasing its win rate to 43%. Now, it's trying to play well, not emulate a weak player.
I've left OpenAI. I'm mostly taking some time to rest. But I also have a few projects in the oven 🧑🍳 Here's one that I'm really excited about: we have a 🚨new paper🚨 out on aligning AI with human values, with the folk at @meaningaligned!! 😊✨🎉 Why I think it's cool: 🧵
“What are human values, and how do we align to them?” Very excited to release our new paper on values alignment, co-authored with @ryan_t_lowe and funded by @OpenAI. 📝: meaningalignment.org/values-and-ali…
This is a very neat idea taking advantage of a feature that AI agents have that humans do not - the ability to evaluate a situation without retaining memory of what it “saw.” AI agents can reveal private information to each other to reduce asymmetry. arxiv.org/pdf/2403.14443…
Jim Halpert retired from the paper industry and working on AI! More seriously though, the most insightful episode of AI from the real frontline🫡
One of the best parts of SF is hanging out with my good friends @dwarkesh_sp and @TrentonBricken. Dwarkesh is the best interviewer in the world - and I hope this gives you a good feeling for what’s it’s like to be on the ground in the labs. It only gets crazier from here!
We finally released the data from our work on evaluating and calibrating AI models with uncertain ground truth. This is an excellent and realistic new benchmark for machine learning with annotator disagreement - a short thread 🧵: github.com/google-deepmin…