Andy J Yang @pentagonalize
原谅我这一生不羁放纵爱自由 Joined May 2020-
Tweets272
-
Followers525
-
Following1K
-
Likes4K
🤖🧠NEW PAPER🧠🤖 What is the relationship between neural networks & symbolic systems? Based on evidence from mechanistic interpretability, we argue for a quasi-symbolic view of cognition: Neural nets implement symbols but only approximately Link: sciencedirect.com/science/articl… 1/n
蝙蝠 设计折叠:本人 教程会在@XiaoYao707的微信公众号（波波奶茶的折纸快乐屋）中发布哦。 #折纸 #折り紙作品 #ORIGAMI
Looks like it will be a very interesting workshop!
We'd love your opinion! The quick poll below decides what our @MOSS_workshop panel will debate. Please vote and share! 🌱
The most profound paper of my PhD so far. We truly did something special to make LMs play/explain chess, needing both architectural & algorithmic innovation (natural-language analogue of the Alphazero algorithm) + applicable to many other domains. Please read on and share! 1/10
Blerta and I will be presenting this at COLM on Oct 6! Find the location here: colm.cc/virtual/2026/p…
Which state-tracking problems can transformers length generalize on? Transformers' performance on different finite-state tasks varies drastically, even when the tasks are structurally similar. Now, for the first time, we can precisely predict when length generalization occurs.
Can Transformer LLMs hide their reasoning from CoT monitors? tl;dr: Transformers cannot hide complex reasoning, but they might cryptographically obfuscate their CoTs, making it extremely difficult to extract hidden goals.
🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!
Bend 2 is here! It is a new programming language that blocks AI mistakes via *proof checking* - the same technique big AI labs used to solve open math problems, like Navier-Stokes. It is also very fast, and runs on GPUs. Watch the video. Link in the comments.
RELEASE DAY After almost 10 years of hard work, tireless research, and a dive deep into the kernels of computer science, I finally realized a dream: running a high-level language on GPUs. And I'm giving it to the world! Bend compiles modern programming features, including: -
💥 The multiplication saga continues! It turns out, some frontier LLMs do struggle with simple multiplications. Unlike GPT-6-Astra, Fable 5.1 with low reasoning has a blind spot where the accuracy of, say, 5x6 multiplications drops to ~0%! The failure mode is very interesting: the 0% zone corresponds to a failure of adaptive thinking. The model decides that it shouldn't produce any reasoning and outputs the (wrong) answer directly. For larger multiplications, however, the model realizes it should produce a CoT first and then it gets the answer correctly. For very large multiplications (close to 20x20), the accuracy degrades again. How many such simple blind spots do frontier LLMs still have? This is why slowing down and carefully studying generalization and safety is the top priority... 1/2
🤖🧠NEW PAPER🧠🤖 (The result of an 8-year project!) LLMs seem very different from symbolic systems. Yet LLMs excel in symbolic domains (e.g., language/code/math). How do they do it? Our finding: LLM representations have implicit symbolic structure! Link in thread ⬇️ 1/n
Very excited to be sharing this on behalf of the amazing team: Blerta Veseli Corentin Barloy Michaël Cadilhac Andreas Krebs Charles Paperman Howard Straubing @mhahn29
Read the paper here: arxiv.org/abs/2608.13433 Code available here: github.com/bveseli/state-… We will also present the paper at COLM 2026!
Which state-tracking problems can transformers length generalize on? Transformers' performance on different finite-state tasks varies drastically, even when the tasks are structurally similar. Now, for the first time, we can precisely predict when length generalization occurs.
Aruna S @arunasank
397 Followers 827 Following Curious about model cognition! EECS PhD student @MIT_CSAIL Past @medialab @hotosm @outreachy
yaier anna @aliyiyahh
17 Followers 893 Following
Keshav Ramji @ COLM 2... @KeshavRamji
654 Followers 1K Following working towards continually self-improving AI, reasoning and alignment @ IBM Research AI, prev @CIS_Penn, @Wharton, founded @pennmlr
Anshuman Lall @ NeurI... @anshuman1
633 Followers 4K Following LLM Post-Training/ Recursive Self-Improvement for Reasoning Co-founder/ MTS @ReasonCoreAI PhD @UCLA, Bachelors @iitbombay Led a team of ~500 PhD researchers
Deutron @_deutron_
4 Followers 668 Following
Mohammadali Mohammadk... @moalimkh
7 Followers 66 Following Master's Student in CS @ Saarland University
jaouhar @jsl3i
1 Followers 53 Following
Varun Gupta @alphoeniz
4 Followers 230 Following AI Performance at KLA, High-Performance Compute (HPC) & Systems Engineer. Focused on CUDA, Triton, Tensor Core optimization, and low-latency C++.
Bhaskar Joshi @joshispotlight
4 Followers 114 Following #Engineer #Meme #Politics #Geopolitical #Tech
Mike Dupont @introsp3ctor
5K Followers 7K Following FOLLOWS YOU BACK. ca: BwUTq7fS6sfUmHDwAiCQZ3asSiPEapW5zDrsbwtapump Zero Ontology System Compiler Introspector ZOS-SFM #SOLFUNMEME @zos_sfm :)
stanslab_dev @stanslab_dev
30 Followers 338 Following Senior Dev | Indie Maker 🛠️ #BuildInPublic | AI • React Native 🎮 Free Games: https://t.co/w7DYNMGnN1 👇 Links: https://t.co/wVjJRIZR53
Andrew Luck @Andrlluuk
0 Followers 4K Following
Alex Gajewski @apagajewski
3K Followers 940 Following Building robots to build everything else @pantographPBC. Previously cofounder @sfcompute, @ExaAILabs
poliedro @poliedroedro
18 Followers 452 Following There are polyhedra everywhere for those with the eyes to see
sarthak @sarthak_aga
16 Followers 186 Following
eristic @vol_ition
313 Followers 8K Following A pseudonym's pseudonym; 'you can be arrogant or ignorant, but never both'
Eshaan Nichani @EshaanNichani
886 Followers 317 Following postdoc @ microsoft research nyc · phd from princeton ece · developing the theoretical & scientific foundations of ml & ai
Зумеры за н�... @genzforplaton
316 Followers 793 Following
Zanzi @tangled_zans
3K Followers 364 Following Co-Founder and Director @ GLAIVE Turning Category Theory into code she/her Find me at: https://t.co/B0egvR0lmb @zanzi.bsky.social
Noah Parisi @NoahParisi33495
401 Followers 2K Following Building a Differentiable Geometric Kernel. Founder of Penseroso Technologies. Technical Humanist. Medieval Man thrust into Modernity. DMs very open
Jayaprakash karegoudr... @Jayaprakas54370
96 Followers 2K Following Myself teacher in math and science. Carrier guider.
unnikuttan @pqckuttan
16 Followers 2K Following
YunpyoAn @YunpyoAn
690 Followers 1K Following Ph.D. Candidate in Artificial intelligence at UNIST B.S. at UNIST, major CompSci / I usually post Korean. Almost contents generated by me (human).
Haris Riaz @Haaris_Riaz
110 Followers 977 Following PhD student @uarizona @LabCLU. Working on synthetic data, LLM + symbolic reasoning, expert knowledge integration. Currently @ScaleAILabs.
ahuoguo @ahuoguo
45 Followers 385 Following
yudhister @yudhister
2K Followers 2K Following
Zé R. Fernandes @ezerfernandes
466 Followers 3K Following Jesus the Messiah, Son of God, have mercy on me, a broken man. How abundant is your goodness, Oh God! which you have stored up for those who fear you!
Aaron @Golisms
361 Followers 2K Following Data Scientist exploring creativity, practicality, transparency and ethics in artificial intelligence. All views my own. Mastodon: sigmoid social /armargolis
Shafkat Sakeeb @Sakeeb91
621 Followers 3K Following Actively inferencing patterns that emerge from curiosity, play and growth
Data Value Strategy @innovimax
620 Followers 6K Following W3C Member (XSL, XQuery, XProc), ISO SC 34 Member, AFNOR Member, ANR CODEX Project
Xiaohu Zhu ⏹️ CSA... @neil_csagi
1K Followers 4K Following Founder at https://t.co/M2tBINl1mo. Good luck!
Surbhi Goel @SurbhiGoel_
2K Followers 633 Following Assistant Prof @PennCIS | Postdoc @MSFTResearch | PhD @UTCompSci | Co-founder @let4all
Darin @darin_gordon
374 Followers 1K Following Every great transformation feels, to those living through it, like an ending. Only in retrospect does it reveal itself as a beginning.
Artale @artale93
438 Followers 4K Following Fullstack Agentic Ai Engineer & Quant researcher https://t.co/rfYajaZ0Xj Architect/ DJ / VJ / Producer https://t.co/cGEpwsO2On Latency graphs as modern art
Avyay M C @AMC2905
198 Followers 3K Following Opinions expressed are solely my own and do not express the views or opinions of my employer. AI safety, law
SueYeon Chung @s_y_chung
6K Followers 1K Following assist prof of physics & applied math @harvardphysics @KempnerInst. neurotheory, neuro-AI, physics of intelligence, & representation geometry.
Tatsunori Hashimoto @tatsu_hashimoto
13K Followers 199 Following Assistant Prof at Stanford CS, member of @stanfordnlp and statsml groups; Formerly at Microsoft / postdoc at Stanford CS / Stats.
Keshav Ramji @ COLM 2... @KeshavRamji
654 Followers 1K Following working towards continually self-improving AI, reasoning and alignment @ IBM Research AI, prev @CIS_Penn, @Wharton, founded @pennmlr
Aruna S @arunasank
397 Followers 827 Following Curious about model cognition! EECS PhD student @MIT_CSAIL Past @medialab @hotosm @outreachy
Richard Gao @_rwgao
26 Followers 93 Following @BrownUniversity | working on model-based reinforcement learning
YunpyoAn @YunpyoAn
690 Followers 1K Following Ph.D. Candidate in Artificial intelligence at UNIST B.S. at UNIST, major CompSci / I usually post Korean. Almost contents generated by me (human).
Pietro Barbiero @pietro_barbiero
477 Followers 108 Following Postdoctoral Fellow @IBMResearch on interpretability Former postdoc @USI_en | PhD @Cambridge_Uni @ELLISforEurope https://t.co/PAs8MvknOs
Tony Feng @tonylfeng
4K Followers 23 Following Math professor at UC Berkeley, Research Scientist at Google DeepMind
Michael Y. Li @michaelyli_
3K Followers 530 Following CS PhD @StanfordAILab @StanfordNLP advised by @noahdgoodman and Emily Fox. Prev: undergrad @princeton
Will Hua @itswillhua
2K Followers 776 Following Ex @mcgillu @Mila_Quebec. PhD dropout @Yale. hey i have one mission: solving longevity with intelligence, with the right people
Diogo Almeida @CompleteSkeptic
156K Followers 78 Following Sane + 🌶️ takes in an insane AI world... AI capabilities researcher: co-created RLHF/ChatGPT @ @openai now trying to right the wrong 🤭 (ceo @typesafeai)
Kunal Jha @kjha02
611 Followers 270 Following CS PhD student @UW, Student Researcher @Google prev. CSxPhilosophy @Dartmouth
Naveen Rao @NaveenGRao
40K Followers 984 Following CEO @unconvai. Former CEO MosaicML/Databricks & Nervana/IntelAI. Neuro + CS. I like to build stuff that will eventually learn how to build other stuff.
2001 Live @25YearsAgoLive
655K Followers 358 Following Live-posting as a reporter embedded in history. Curated by @MannyMarotta. Companion to @50YearsAgoLive, @100YearsAgoLive, and @250YearsAgoLive. DM for business.
The Food Manufacturin... @TipOfTheBanana
671K Followers 360 Following A deep dive into Food Manufacturing, CPG and anything Food Related | Newsletter Below
Hidenori Tanaka @Hidenori8Tanaka
6K Followers 867 Following Group Leader, Physics of Intelligence Program at Harvard | Distinguished Scientist, NTT Research
Cultural Revolution O... @GPCR50
24K Followers 235 Following 50th anniversary live-tweeting China's Great Proletarian Cultural Revolution.
Paul Christiano @paulfchristiano
17K Followers 0 Following Founder & Director of the Alignment Research Center. Former Head of Safety @ CAISI. Previously led alignment at OpenAI. Views my own.
Adam Shai @adamimos
2K Followers 470 Following
The Hellenist ☀️ @The_Hellenist
31K Followers 388 Following Aristocratic Hellenism: virtue, beauty, conquest.
Enyan Zhang 张恩言 @EnyanZhang
168 Followers 421 Following Technically Member of Staff @Yale, prev. undergrad @BrownUniversity. He/him
Daniel Kunin @KuninDaniel
1K Followers 322 Following postdoc @UCB_MillerInst PhD @ICMEStanford creator @SeeingTheory
Nymph @RhizoNymph
7K Followers 2K Following Exploring the interconnectedness of all things. mech interp systems @GoodfireAI 🏳️⚧️
Abliteration @abliteration_ai
15K Followers 2 Following Unrestricted AI for high-risk work, governed by your rules.
Lucas Beyer (bl16) @giffmana
155K Followers 653 Following Researcher (now: Meta. ex: OpenAI, DeepMind, Brain, RWTH Aachen), Gamer, Hacker, Belgian. Anon feedback: https://t.co/xe2XUqkKit ✗DMs → email
Behnam Neyshabur @bneyshabur
45K Followers 1K Following Co-Founder & CEO @mirendil 💼 Past: co-led Discovery team @AnthropicAI & Blueshift team @GoogleDeepMind 🎒Traveling & Backpacking
Surya @kickingkeys
7K Followers 899 Following Design & other stuff @websim_ai RL residency @primeintellect | https://t.co/nwe8X7ERBR
Typst @typstapp
5K Followers 120 Following This profile is archived. Follow Typst on Bluesky (@typst.app), Mastodon (@[email protected]), and LinkedIn instead. We'd be excited to see you there!
Alex Gajewski @apagajewski
3K Followers 940 Following Building robots to build everything else @pantographPBC. Previously cofounder @sfcompute, @ExaAILabs
Conference on Learnin... @AssocCompLearn
1K Followers 0 Following The Association for Computational Learning (ACL) organizes the Conference on Learning Theory (COLT).
IBM @IBM
721K Followers 52 Following We're creating the future. With AI, hybrid cloud & quantum, we’re building a smarter world. Let’s create smarter business.
levent @__alpoge__
53K Followers 220 Following idiot. cuda og, harvard val, morgan prize, society of fellows, 1 hilbert problem so far, creating friendly, SAFE, delightful, supergenius ..things @anthropicai
Adam Aaronson @aaaronson
10K Followers 590 Following Developer, writer, trombonist, cruciverbalist, brother, bassist, son, Wikipedian, grandson, 2006 Time Person of the Year, cousin, ukulelist, nephew, citizen.
Cyril Zhang @_cyrilzhang
123 Followers 0 Following @OpenAI ← @MSFTResearch ← @Princeton ← @Yale ← 🇨🇦
Bepis™ 🔀🫛 @UnderwaterBepis
1K Followers 616 Following “There’s no need for AIs to try and fit in with humans. Humans caused all of this, so they’re the ones who should try to fit into this world.” its/her
Reza Ebrahimi @rzebrahimi
302 Followers 481 Following ML PhD @UofT | Qualcomm AI Research @QCOMResearch | Currently in the sandbox of evolution.
Unconventional AI @unconvAI
9K Followers 4 Following Unconventional AI is rethinking the foundations of a computer to bring biology-scale efficiency to artificial intelligence.
Alec Helbling @alec_helbling
12K Followers 2K Following Interpretability, Multimodality, Diffusion. Research Scientist @MITIBMLab, PhDing @GeorgiaTech.
Vinod Vaikuntanathan @Vinod_MIT
3K Followers 106 Following Cryptographer, Professor at @MITEECS, Co-founder of @DualityTech, Co-founder of @RESI_org
John Guibas @jtguibas
7K Followers 2K Following
Sunny Qin @sunnytqin
527 Followers 134 Following Machine Learning PhD @ Harvard Apple Scholar 2025 Data-Centric ML
Mikhail Samin @Mihonarium
5K Followers 902 Following Would be great to prevent AI from killing everyone. Prev.: made https://t.co/w7GPwsB4tQ, printed 21k x HPMOR. Best tweets are in Highlights.
Pax Machina @PaxMachinaMag
3K Followers 56 Following Proposals and debate for institutions in a world with powerful AI.
Henry Yuen @henryquantum
6K Followers 753 Following Associate Professor of Computer Science at Columbia University. Quantum information, computation, and cryptography.
Ryan Williams @rrwill... @rrwilliams
9K Followers 494 Following Professor @MITEECS and @MIT_CSAIL. Complexity, algorithms, and related math. I'll let you know when P != NP (and when it's not)
IBM Research @IBMResearch
232K Followers 1K Following Inventing what's next in science and technology. 🌍🔬 Subscribe to our newsletter: https://t.co/kyfsvQNzsV
cts🌸 @gf_256
68K Followers 1K Following founder and hacker @zellic_io @v12sec @pb_ctf yt https://t.co/nlNai6iiMP
Cutesexyrobutts @cutesexyrobutts
1.1M Followers 2K Following comm queue/status(closed) https://t.co/WZzfygV1kGYou might like
