Taylor Webb (@taylorwwebb.bsky.social) @TaylorWWebb
Studying cognition in humans and machines. taylorwwebb.github.io Joined October 2017-
Tweets321
-
Followers609
-
Following458
-
Likes465
In-context learning (ICL) circuits emerge in a phase change... Excited for our new work "What needs to go right for an induction head (IH)?" We present "clamping", a method to causally intervene on dynamics, and use it to shed light on IH diversity + formation. Read on 🔎⏬
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)!
⚠️PREPRINT⚠️ My first paper! Do you know everything about DNNs, human vision and recurrent processing? No? We neither. But we tried to learn a bit about them, and our results are not what we expected. Check out our work with @hans_opdebeeck & @costantino_ai ⬇️⬇️ 1/9
New preprint w/ @mcxfrank: bit.ly/3VMdl2u How can we ascribe cognitive abilities to language models? We evaluate them! But evals impose challenges separate from the underlying ability of interest. These "task demands" affect LM performance, esp. for smaller models! 1/8
Very provocative! Looking forward to reading this one.
i plan to move to Korea later this year, & will soon hire at all levels (students, postdocs, staff scientists, junior PIs) - docs.google.com/document/d/1Vw… my lab in Japan will run till at least 2025. RIKEN has been nothing but the dream job. but hopefully the above explains it. 1/2
I am hoping to hire a postdoc who would start in Fall 2024. If you are interested in the intersection of linguistics, cognitive science, and AI, I encourage you to apply! Please see this link for details: rtmccoy.com/prospective_po…
From task structures to world models: what do LLMs know? Opinion by Ilker Yildirim & L.A. Paul bit.ly/43auhRR
Out today in Neuron 🥳 doi.org/10.1016/j.neur… ⚖️ Is there a computational trade-off between memory and abstraction? 📐 How does the geometry of the neural code control this trade-off? 🧠 What are its implications on social memory and familiarity? Check out the paper and 🧵👇
📢 📢 Delighted to share our paper with the amazing team Safoora Yousefi, Leo Betthauser, @grockious @raphaelmilliere on neuro-inspired analysis of LLMs. Decoding In-Context Learning: Neuroscience-inspired Analysis of Representations in LLMs arxiv.org/abs/2310.00313 🧵1/n #Llama
New paper alert! There's a wealth of evidence that infants categorize spatial relations starting from 3-4 months onward. Can we evaluate neural networks in similar paradigms to babies, without special relational training? If so, what do we learn? 1/N sciencedirect.com/science/articl…
It is an amazing time to work in the cognitive science of language. Here are a few remarkable recent results, many of which highlight ways in which the critiques of LLMs (especially from generative linguistics!) have totally fallen to pieces.
My article on how apical dendrites of pyramidal neurons can implement a computation similar to transformer attention is now published in PLOS Computational Biology journals.plos.org/ploscompbiol/a…
Can insights from synthetic experiments and interpretability lead to real improvements in language modeling? We: > propose a formal model for in-context learning > uncover "n-gram heads" = high order induction heads, crucial for ICLL > improve Transformer LM perplexity by 6.7%
> Transformers significantly outperform neural sequence models with recurrent or convolutional representations on ICLL tasks […] we provide evidence that their ability to do so relies on specialized “n-gram heads” (higher-order variants of previously-described “induction heads”)
Now out in Cognition, @yanlisagezhou models how people learn compositional visual concepts through both Bayesian program induction and meta-learning. There are two stories in her work... (thread below) sciencedirect.com/science/articl…
People don't take a one-size-fits-all approach to planning: we change how we abstract the world to fit our goals. Our approach, Ada, integrates LLMs + formal planning to learn libraries of composable skills adapted to individual planning domains: tdy.lol/BXnzn 🧵 [1/5]
New preprint w/ @callfredaway @cocosci_lab osf.io/preprints/psya… We show that simulated data from cognitive models can be used to pretrain neural networks to infer an individual's preferences from their fixations.
Can Transformers be used to learn object-centric representations? Yes! If we "invert" the attention operation by taking the softmax over the queries instead of the keys. paper: openreview.net/pdf?id=m9s6rnY…
@kenshin9000_ Personally, I am optimistic about LLMs and what they might lead to (even AGI, maybe), but this is confusing. I looked at the prompts. You are giving a different prompt for each problem. In the extreme, one might as well give the solution directly and ask GPT to copy it 1/2
Jorge Morales (@jorge.. @jorgemlg
4K Followers 3K Following I study visual awareness & introspection. Assistant professor of Psychology & Philosophy @Northeastern Find me at: https://t.co/bcxxd68JXWSteve Fleming @smfleming
12K Followers 2K Following Professor of Cognitive Neuroscience @ucl | Author #KnowThyself | Trying to keep up with @HWalkerFleming 🇬🇧🇭🇷Marion Rouault @marion_rouault
2K Followers 2K Following Cognitive neuroscientist at @CNRS and @InstitutCerveau interested in decision-making, metacognition and mental health https://t.co/TuKqKsFCY3Ida Momennejad @criticalneuro
15K Followers 2K Following Principal Researcher @MSFTResearch. I study memory/planning in brains. I build/evaluate AI. Writing on Memory. Finishing a novel. Cohost https://t.co/rgvLjhBvSHmegan peters 🧠 on .. @meganakpeters
5K Followers 2K Following assoc prof @ucicogsci: perception+confidence+consciousness, neuroimaging+modeling+ML. prez+co-founder @neuromatch. fellow @cifar_news. she/her. views mine.Melanie Mitchell @MelMitchell1
44K Followers 657 Following Professor, Santa Fe Institute. More thoughts at https://t.co/nC43NHRozX.Mark Ho @mark_ho_
3K Followers 2K Following Computational Cognitive Scientist • I’m interested in human problem solving and social cognition • Asst Prof @FollowStevens • he/they • https://t.co/gdqtrgIGOAhakwan lau 🇺🇦 @.. @hakwanlau
8K Followers 3K Following PI@Riken Center for Brain Science, Japan; HongKonger // perceptual metacognition: implicit & explicitAngela Radulescu @angelaradulescu
2K Followers 666 Following assistant prof @MountSinaiPsych, PI @sinaiCCP | studying cognitive and affective dynamics in mental health | 🌈 (they/she)Pete Mandik @petemandik
6K Followers 1K Following Freestanding utility protein. https://t.co/apJUV1Do1JRobert Hawkins @hawkrobe
3K Followers 1K Following computational cognitive scientist @UWPsych. he/they. https://t.co/TI4yjMJIfR.Laura Ana Bustamante,.. @lauraabusta
2K Followers 3K Following postdoc @WUSTL researching cognitive control, #neurodivergent in computational psychiatry, cuban-american she/her, find me on 🟦☁️Sreejan Kumar @sreejan_kumar
2K Followers 305 Following PhD candidate at Princeton @cocosci_lab. Yale '19.Jesse Rissman @jesse_rissman
2K Followers 959 Following UCLA cognitive neuroscientist, explorer of memories old and new.summerfieldlab @summe.. @summerfieldlab
8K Followers 811 Following Investigating the mechanisms that underpin human learning, perception and cognition, headed by Chris SummerfieldMatthias Michel @MatthiasMichel_
4K Followers 719 Following Philosophy of science and cognitive science of consciousness. Center for Mind, Brain and Consciousness, New York University.Harrison Ritz @harrison_ritz
2K Followers 3K Following cybernetic cognitive control 🤖 computational cognitive neuroscience 🧠 postdoc princeton neuro 🍕 he/him 🇨🇦Curtain @Curtain1923435
3 Followers 190 FollowingMars Bay Studio @bazaerbaike
734 Followers 4K Following Embark on a journey with Mars Bay Studio through stunning 4K AI videos and a symphony of sounds spanning genres. Subscribe for exquisite visuals and tunes.ben @ben_woodward
321 Followers 312 FollowingAdele Goldberg @adelegoldberg1
7K Followers 2K Following Linguist, experimentalist, constructionist usage-based approach to language (she's)Pensé FFun @inftyCategory
96 Followers 6K FollowingZohaib Hassan, PhD @drzohaibh
889 Followers 6K Following Postdoctoral Researcher at @ZhejiangLab HangzhouColton Casto @_coltoncasto
111 Followers 326 Following PhD student @Harvard @MIT - neuroscience, language, AIDorian Liu @LiuZT612
47 Followers 229 FollowingEunice Yiu @eunice_yiu_
30 Followers 6 Following PhD student at @BerkeleyPsych advised by @AlisonGopnikNicolò De Sabbata @cndesabbata
99 Followers 439 Following 👨🏻💻CS MS student at @EPFL_en | Visiting Student Researcher @Princeton |Previously @AXA @amazon | 🧠Deep Learning, NLP & Cognitive Science | 🇪🇺🇮🇹🇨🇭wishnu @wishnuss
2 Followers 46 FollowingBonan Zhao / 赵博�.. @BonanZhao
409 Followers 282 Following CoCoSci postdoc at Princeton (she/her). Computational models, concepts, discovery, causality & generalization.Ziyi Yang @ZiyiYang96
39 Followers 176 Following CS PhD student at @BrownCSDept; Robotics + LanguageMichael Pascale @mpascale00
178 Followers 715 Following Yr1 PhD Student, Boston University Brain Behavior Cognition he/him 🇺🇸 🥼 🤝 🧠 🎓 🇷 💾 🌈 Human Information Behavior | Formerly @MGHPsychiatry & @UMassLowellMichael Malina @MikeyMalina
36 Followers 231 Following PhD Student, Systems Neuroscience, studying sensory guided behavior, Carnegie Mellon UniversityRicky W.J. Choi @RickyWJChoi
79 Followers 272 Following “I will not 사과 for my love for 사과“ 🍎 Developmental cognitive scientist in training 💭 Swarthmore '23 | PhD student @cmupsychWai Keen Vong @wkvong
1K Followers 1K Following research scientist in computational cognitive science @nyuniversity. interested in concepts, language and abstractionAmir Arsalan Soltani @Amir_Arsalan_
230 Followers 905 Following In pursuit of the Truth without fear of asking questions in this judgemental world.RTs/sharing/likes/comments are result of my struggle for this,not endorsementSchacter Memory Lab @SchacterLab
268 Followers 147 Following Our research explores human memory - impacts of aging, the nature of memory distortions & how individuals use memory to imagine possible future eventsYarrow Dunham @yarrowdunham
3K Followers 2K Following Studies social cognition in children and grown-ups. Teaches in the psychology and cognitive science programs at Yale.yuqing l @ryannlei
1 Followers 56 FollowingKyle Marieb @kylemarieb
742 Followers 5K Following Profoundly deaf with cochlear implants 🦻🤖 YouTube Backend SWE 📺Awni Altabaa @Awni00
2 Followers 60 Following PhD Student @Yale Stats. Interested in statistical machine learning, reinforcement learning, and the neuro/AI interface. (trying out "academic twitter" 🤷)Phoebe Friesen @PhoebeFriesen
1K Followers 1K Following Philosopher thinking about research ethics/ medical ethics /critical psychiatry/ feminist philosophy of science/ the placebo effect (she/her)ruchira sewmini @RSewmini67609
13 Followers 125 FollowingPoornima Lakshani @PoorniLK
5 Followers 70 FollowingAmila @amila_wjsr
115 Followers 1K Following e/acc, Neural Nerd, Techno-Optimist, AI AccelerationistDaniel @drag0n123432
127 Followers 377 FollowingAndra Mihali @Lianaaan
523 Followers 2K FollowingSixuan Chen @iamcsx
59 Followers 277 Following MA Cognitive Neuroscience at University of Waterloo, BA Neuroscience CS/Math Minors at Colorado CollegeMichael Lepori @Michael_Lepori
299 Followers 479 Following CS PhD Student at Brown interested in the intersection of machine learning and cognitive science, but more interested in playing guitar. He/Him.Doris Yu @DorisYu00
13 Followers 93 FollowingElley @elleyccn
5 Followers 164 Followingnora bradford @noraabradford
804 Followers 861 Following @norabradford on bluesky cog sci PhD student @ucirvine w/Megan Peters science writer studied phil + neuro + psych @uchicago she/they🥀 @PsyNetMessage
477 Followers 3K FollowingLeastoa @leastoa72055
119 Followers 4K FollowingWarrenGMartin @WarrenGMartin
18 Followers 60 FollowingIsaac Thompson @IsaacThss
111 Followers 1K Following Research Assistant at The Alan Turing Institute, researching reinforcement learning and autonomous cyber defence.Layton Lamsam @Layton_Lamsam
23 Followers 123 FollowingZiqing Shi (Seesaw) @ZiqingShi
504 Followers 4K Following Neurology PostDoc @Yale I like Mice Behavior/Microcontrollers/Sensors/RoboticsMola 相羊 @xiangya94910377
55 Followers 2K Following a psychology student ,curious about neural decoding、cognitive mathematics 、emergent communicationShi Gu @gushilab
324 Followers 322 Following On the academic job market 2024, looking for tenured /tenure positions. Professor @ UESTC Network Neuroscience & Artificial Intelligencecarlos g. correa @_cgcorrea
668 Followers 3K Following phd candidate @princetonneuro, thinking about decisionsJinwoo Yi @aesciemo
114 Followers 518 Following MS student in Neuroscience, BA in Psychology & Aesthetics / study affect dynamics in aesthetic experience and its relationship with resilienceJessica282931 @jessica282931
16 Followers 875 FollowingYann LeCun @ylecun
712K Followers 719 Following Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.Jorge Morales (@jorge.. @jorgemlg
4K Followers 3K Following I study visual awareness & introspection. Assistant professor of Psychology & Philosophy @Northeastern Find me at: https://t.co/bcxxd68JXWSteve Fleming @smfleming
12K Followers 2K Following Professor of Cognitive Neuroscience @ucl | Author #KnowThyself | Trying to keep up with @HWalkerFleming 🇬🇧🇭🇷Marion Rouault @marion_rouault
2K Followers 2K Following Cognitive neuroscientist at @CNRS and @InstitutCerveau interested in decision-making, metacognition and mental health https://t.co/TuKqKsFCY3David Chalmers @davidchalmers42
29K Followers 667 Following philosopher@NYU. consciousness, reality+, life, the universe, and everything.Ida Momennejad @criticalneuro
15K Followers 2K Following Principal Researcher @MSFTResearch. I study memory/planning in brains. I build/evaluate AI. Writing on Memory. Finishing a novel. Cohost https://t.co/rgvLjhBvSHmegan peters 🧠 on .. @meganakpeters
5K Followers 2K Following assoc prof @ucicogsci: perception+confidence+consciousness, neuroimaging+modeling+ML. prez+co-founder @neuromatch. fellow @cifar_news. she/her. views mine.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.Melanie Mitchell @MelMitchell1
44K Followers 657 Following Professor, Santa Fe Institute. More thoughts at https://t.co/nC43NHRozX.Mark Ho @mark_ho_
3K Followers 2K Following Computational Cognitive Scientist • I’m interested in human problem solving and social cognition • Asst Prof @FollowStevens • he/they • https://t.co/gdqtrgIGOADr. Anna Schapiro @AnnaSchapiro
9K Followers 827 Following Assistant Prof at U Penn. Learning, memory, sleep, neural network modeling, open science. https://t.co/4dODyQuOd0Joseph E. LeDoux "The.. @theamygdaloid
14K Followers 1K Following @[email protected] Neuroscientist at NYU and member of the rock band The Amygdaloids.hakwan lau 🇺🇦 @.. @hakwanlau
8K Followers 3K Following PI@Riken Center for Brain Science, Japan; HongKonger // perceptual metacognition: implicit & explicitAngela Radulescu @angelaradulescu
2K Followers 666 Following assistant prof @MountSinaiPsych, PI @sinaiCCP | studying cognitive and affective dynamics in mental health | 🌈 (they/she)AI at Meta @AIatMeta
532K Followers 255 Following Together with the AI community, we are pushing the boundaries of what’s possible through open science to create a more connected world.Pete Mandik @petemandik
6K Followers 1K Following Freestanding utility protein. https://t.co/apJUV1Do1JColton Casto @_coltoncasto
111 Followers 326 Following PhD student @Harvard @MIT - neuroscience, language, AIMichael C. Frank @mcxfrank
15K Followers 2K Following Cognitive scientist at Stanford. Open science advocate. @stanfordsymsys director. Bluegrass picker, slow runner, dad.Kanishk Gandhi @gandhikanishk
920 Followers 692 Following Phd @Stanford CS; w/ Noah Goodman, Dorsa Sadigh | Prev: @LakeBrenden @NYUDataScience, @IITKanpur, @Path_AIJessica Cantlon @CantlonLab
6K Followers 2K Following Professor of Psychology studying origins of logic @CarnegieMellon; Brain development, evolution; Primate Portal @SenecaParkZoo: @TIME 2017 Silence Breakers; sheWayne Wu @attninaction
999 Followers 77 Following CMU Philosophy & Neuroscience Institute Editor WIREs Cogsci https://t.co/ZatUTANpP8… Author of https://t.co/u99YFF0imt https://t.co/yEcJVqs8hSBarbara Pomiechowska .. @B_Pomiechowska
689 Followers 1K Following Cognitive neuroscientist & developmental psychologist, studies how developing minds and brains become creative, Asst Prof @UoB_SoP @TheCHBH @UoB_CDSEunice Yiu @eunice_yiu_
30 Followers 6 Following PhD student at @BerkeleyPsych advised by @AlisonGopnikM.J. Crockett @mollycrockett
15K Followers 2K Following Professor @PsychPrinceton & University Center for Human Values | Cognitive scientist curious about (anti)normativity, technology & the self | They/She 🏳️🌈Bonan Zhao / 赵博�.. @BonanZhao
409 Followers 282 Following CoCoSci postdoc at Princeton (she/her). Computational models, concepts, discovery, causality & generalization.Ziyi Yang @ZiyiYang96
39 Followers 176 Following CS PhD student at @BrownCSDept; Robotics + LanguageEliza Kosoy @ElizaKosoy
287 Followers 350 Following Ph.D. Student at @UCBerkeley/@berkeley_ai working in child development and AI with @AlisonGopnik AI/UX @google @elizadolls Prev: @mitcocosci Ukrainian-AmericanMichael Pascale @mpascale00
178 Followers 715 Following Yr1 PhD Student, Boston University Brain Behavior Cognition he/him 🇺🇸 🥼 🤝 🧠 🎓 🇷 💾 🌈 Human Information Behavior | Formerly @MGHPsychiatry & @UMassLowellMichael Malina @MikeyMalina
36 Followers 231 Following PhD Student, Systems Neuroscience, studying sensory guided behavior, Carnegie Mellon UniversityRicky W.J. Choi @RickyWJChoi
79 Followers 272 Following “I will not 사과 for my love for 사과“ 🍎 Developmental cognitive scientist in training 💭 Swarthmore '23 | PhD student @cmupsychSchacter Memory Lab @SchacterLab
268 Followers 147 Following Our research explores human memory - impacts of aging, the nature of memory distortions & how individuals use memory to imagine possible future eventsWai Keen Vong @wkvong
1K Followers 1K Following research scientist in computational cognitive science @nyuniversity. interested in concepts, language and abstractionWill Koller @WilliamNKoller
428 Followers 615 Following Clinical psychology Ph.D. candidate @Yale studying psychosis, psychotic-like experiences, & the memory/learning processes that may underlie them.Yarrow Dunham @yarrowdunham
3K Followers 2K Following Studies social cognition in children and grown-ups. Teaches in the psychology and cognitive science programs at Yale.Chris Olah @ch402
91K Followers 173 Following Reverse engineering neural networks at @AnthropicAI. DMs open! Previously @distillpub, OpenAI Clarity Team, Google Brain. Personal account.Ekin Akyürek @akyurekekin
2K Followers 727 Following graduate student in computer science @MITEECS/@MIT_CSAILTianmin Shu @tianminshu
883 Followers 394 Following assistant professor @JHUCompSci & @JHUCogSci | working on social AI, embodied AI, and computational social cognitionPhoebe Friesen @PhoebeFriesen
1K Followers 1K Following Philosopher thinking about research ethics/ medical ethics /critical psychiatry/ feminist philosophy of science/ the placebo effect (she/her)François Fleuret @francoisfleuret
31K Followers 456 Following Prof. @Unige_en, Adjunct Prof. @EPFL_en, Research Fellow @idiap_ch, co-founder @nc_shape. AI and machine learning since 1994. I like reality.Ashish Vaswani @ashVaswani
19K Followers 2K FollowingMichael Lepori @Michael_Lepori
299 Followers 479 Following CS PhD Student at Brown interested in the intersection of machine learning and cognitive science, but more interested in playing guitar. He/Him.Kayuet Liu @KayuetLiu
59 Followers 56 Following Associate Professor of Sociology at UCLA. Senior Staff Scientist at RIKEN, Japan.nora bradford @noraabradford
804 Followers 861 Following @norabradford on bluesky cog sci PhD student @ucirvine w/Megan Peters science writer studied phil + neuro + psych @uchicago she/theyZiqing Shi (Seesaw) @ZiqingShi
504 Followers 4K Following Neurology PostDoc @Yale I like Mice Behavior/Microcontrollers/Sensors/RoboticsLayton Lamsam @Layton_Lamsam
23 Followers 123 FollowingShi Gu @gushilab
324 Followers 322 Following On the academic job market 2024, looking for tenured /tenure positions. Professor @ UESTC Network Neuroscience & Artificial Intelligencecarlos g. correa @_cgcorrea
668 Followers 3K Following phd candidate @princetonneuro, thinking about decisionsSheridan Feucht @sheridan_feucht
109 Followers 227 Following PhD student @KhouryCollege working on LLM interpretability. Also affiliated with @Brown_NLP as an undergrad ('23)Leyla Isik @lisik.bsk.. @leylaisi
2K Followers 1K Following Cognitive neuroscientist studying visual and social perception. Asst Prof @jhucogsci. She/her.Manasi Malik @ManasiMalik
176 Followers 170 Following Computational Cognitive Science Grad at @JHUCogSci | @IIITDelhi alum (she/her) https://t.co/l24mm5Da1BIsaac Chotiner @IChotiner
111K Followers 1K Following New Yorker staff writer and Houston Rockets fan; formerly a staff writer at Slate and senior editor at The New Republic.Alexander Ku @alex_y_ku
345 Followers 298 Following Research scientist @GoogleDeepMind. PhD student @PsychPrinceton and @PrincetonNeuro.Tiwa Eisape @tiwa_eisape
1K Followers 1K Following PhD student at @MIT working on NLP and cognitive science - @NSF grfp fellow. Previously with @GoogleAI and @Meta FAIRShuze Liu @liu_shuze
28 Followers 231 FollowingShane Shang @shane_shang1024
239 Followers 1K Following Theoretical Neuroscience Ph.D. Student @ Harvard with Prof. Haim Sompolinsky and Prof. Gabriel KreimanWilka Carvalho @cogscikid
1K Followers 763 Following kempner fellow @Harvard studying human reinforcement learning. cognitive scientist & AI researcher. latinx new yorker. prev: umich, deepmind, msrJennifer Hu @_jennhu
2K Followers 97 Following Research Fellow at @Harvard and incoming Asst Prof at @JohnsHopkins interested in language, computation, and cognition. @jennhu.bsky.socialBarbara Lee @BarbaraLeeForCA
72K Followers 1K Following California’s Effective Progressive. Candidate for US Senate. Proud Congresswoman from the East Bay. Text 'LEE' to 43100.Antonino π Greco @_AntoninoGreco
425 Followers 918 Following Computational Cognitive Scientist • NeuroAI, Predictive Coding, Reinforcement & Deep Learning, Complex Systems • Postdoc at @siegull @uni_tue • Husband & DadPranav Mahajan @iam_mahajan
679 Followers 4K Following DPhil student at NDCN, University of Oxford - Computational neuroscientist thinking about pain. (Views are my own, pronouns: he/his)Ted Moskovitz @ted_moskovitz
743 Followers 193 Following PhD student at @GatsbyUCL. Formerly: intern at @DeepMind, @UberAILabs, student at @ColumbiaCompSci, @PrincetonNeuro.Excited to announce our full-day workshop on “In-context learning in natural and artificial intelligence” at CogSci (@cogsci_soc) 2024 in Rotterdam (with @JacquesPesnot @akjagadish @summerfieldlab and Ishita Dasgupta). jacquespesnot.github.io/2024_CogSci_Wo…
Come to Rotterdam and chat about in-context learning with us!
Excited to announce our full-day workshop on “In-context learning in natural and artificial intelligence” at CogSci (@cogsci_soc) 2024 in Rotterdam (with @JacquesPesnot @akjagadish @summerfieldlab and Ishita Dasgupta). jacquespesnot.github.io/2024_CogSci_Wo…
And the language they interpret and produce is a million times better than any symbolic system ever dreamt of
I used to be a skeptic too. Sure, LLMs fall down on math & complex logic (as do lots of humans). They also learn w/o real world contexts. But they understand what's Relevant & respond appropriately! When it comes to lng, they seem to be as creative as humans
@adelegoldberg1 I thought this was a joke. What? You mean someone at Princeton thinks there is any relation between LLMs and human cognition? I can't criticize the misguided (paid?) media anymore. They will reply: hey, it's not only me dude, check out what Princeton is saying.
So honored to be awarded the Glushko CogSci dissertation award, along with 6 amazing scientists! I am grateful to my advisor @alex_ander and wonderful collaborators for shaping my research. See you in the Netherlands for a talk on my dissertation at CogSci2024! 🤖🧠💬
Sean Dae Houlihan @DaeHoulihan – 2022 PhD Thesis “A Computational Framework for Emotion Understanding”, from @MIT Shailee Jain @shaileeejain – 2023 PhD Thesis “Investigating the neurobiology of language with artificial language systems”, from @UTAustin
I am incredibly honored to receive a Glushko Dissertation Prize! A huge thank-you goes to: - My dissertation advisors, @tallinzen and @paul_smolensky, for being incredibly supportive throughout my PhD - (continued in next tweet) 1/2
The Cognitive Science Society is thrilled to announce the winners of the 2024 Glushko Dissertation Prize! 🏆 Let’s meet the brilliant minds behind groundbreaking research in Cognitive Science 🧵👇
I’m super excited to announce that I’ll be starting a PhD in Computer Science at @Stanford this upcoming fall! Stay posted for (hopefully) cool work on visual abstraction, emergence of symbols in DNNs, & cognitively-inspired interpretability 🟦🔴
Postdoc job alert! The Collective Cognition Lab is moving moving to @Stanford in the Department of Environmental Social Sciences @Stanforddoerr & looking for a postdoctoral scholar interested in the psychology of climate beliefs and behaviors. Apply now: forms.gle/smfec81pMR8XzK…
When I moved to California for my PhD, I felt like I was home. I never wanted to leave. So I am *delighted* to be going back – to join the faculty @UCBerkeley! Thanks to my lab, mentors, letter writers, friends, & partner for helping me land a dream job @BerkeleyPsych❤️☀️ (1/3)
Our new paper delves into the circuits and training dynamics of transformer in-context learning (ICL) 🥳 Key highlights include 1️⃣ A new opensourced JAX toolkit that enables causal manipulations throughout training 2️⃣ The toolkit allowed us to "clamp" different subcircuits to…
In-context learning (ICL) circuits emerge in a phase change... Excited for our new work "What needs to go right for an induction head (IH)?" We present "clamping", a method to causally intervene on dynamics, and use it to shed light on IH diversity + formation. Read on 🔎⏬
In-context learning (ICL) circuits emerge in a phase change... Excited for our new work "What needs to go right for an induction head (IH)?" We present "clamping", a method to causally intervene on dynamics, and use it to shed light on IH diversity + formation. Read on 🔎⏬
@TaylorWWebb @MelMitchell1 @_jennhu Agreed! I think the hope that I have for this project is that deeper investigation of the relationship between LLM task demands and task demands for children can shed light on both.
@MelMitchell1 @_jennhu Also as a side note, LLMs *do* have attention, and you could imagine trying to analyze attention maps to understand where particular evaluations fail (e.g., the meta-linguistic prompts we looked at).
@mcxfrank @MelMitchell1 @_jennhu I agree! Ideally we'd also be able to draw on mechanistic evidence to arbitrate these disputes; e.g. evidence that some task/prompt variations fail to trigger an existing circuit that would otherwise enable the model to solve the task.
@MelMitchell1 @_jennhu Thanks Melanie! @_jennhu should say what she thinks. For me, I agree that it's all autoregression in the end, but the task demands analogy can be fruitful by explicitly separating the construct being evaluated from the nature of the evaluation.
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)!
⚠️PREPRINT⚠️ My first paper! Do you know everything about DNNs, human vision and recurrent processing? No? We neither. But we tried to learn a bit about them, and our results are not what we expected. Check out our work with @hans_opdebeeck & @costantino_ai ⬇️⬇️ 1/9
New preprint w/ @mcxfrank: bit.ly/3VMdl2u How can we ascribe cognitive abilities to language models? We evaluate them! But evals impose challenges separate from the underlying ability of interest. These "task demands" affect LM performance, esp. for smaller models! 1/8
Many congratulations Matthias- this is wonderful news! Excellent choice MIT!! 👏👏👏
Very happy to announce that I will soon join the MIT linguistics and philosophy department as an assistant professor!
Our project tries to capture the STEM hearts of elementary school students by letting them write their own code to test our monkeys’ cognition. This new study found that they learn “hard skills” AND grow STEM interest, Open Access: link.springer.com/article/10.100…