mike64_t @mike64_t
descending the gradient Joined October 2022-
Tweets3K
-
Followers4K
-
Following314
-
Likes12K
This is obviously true because there is data that it is arguably impossible to tokenize. Or else we would be throwing LLMs at h264 codec bytes directly.
This is obviously true because there is data that it is arguably impossible to tokenize. Or else we would be throwing LLMs at h264 codec bytes directly.
Two things can be true at the same time CE on next token & RLVR are good objectives. The bottlenecks of Attention will starve us before we will get to AGI.
This is what Frame-perfect keystroke recovery from Gameplay footage looks like.
Imagine not being able to reason about the inductive biases of your model and thus needing a test set (I’m only 50% joking)
Imagine not being able to reason about the inductive biases of your model and thus needing a test set (I’m only 50% joking)
Sometimes the best thing you can do is hardcode weights...
Serious infrastructure always has the last laugh. The principled planner always catches up to the greedy rusher in the long run. Shame on all who were bearish on Mojo two years ago.
Serious infrastructure always has the last laugh. The principled planner always catches up to the greedy rusher in the long run. Shame on all who were bearish on Mojo two years ago.
It just occurred to me how much YUV screws with visuals when the source data is RGB. Also, people who may have missed a time when everyone was using Fraps to record gameplay may have forgotten how good near uncompressed video looks. Filling up your disk with a 1TB FullHD video is…
https://t.co/YWtkPifark
There is a whole category of expressions systematically underexplored because there is this strange notion of what an "architecture" must be. We've been tuning norms and position embeddings to death, slightly modifying attention here and there, giving you the Transformer+++++™.…
There is a whole category of expressions systematically underexplored because there is this strange notion of what an "architecture" must be. We've been tuning norms and position embeddings to death, slightly modifying attention here and there, giving you the Transformer+++++™.…
1/ The myth: 2017 “invented” attention. The reality: content-addressable memory and fast weights (early ’90s), self-referential nets (’93), LSTM for long dependencies (’97), and encoder-decoder attention in NMT (2014). Old wine, new label.
1/ The myth: 2017 “invented” attention. The reality: content-addressable memory and fast weights (early ’90s), self-referential nets (’93), LSTM for long dependencies (’97), and encoder-decoder attention in NMT (2014). Old wine, new label.
A case for open infrastructure
@MatthewJBar Open source infrastructure does not imply that everybody is now suddenly inferencing models on consumer grade hardware for shits and giggles as you seem to be suggesting. Ultimately, accumulating RL envs in a centralized fashion succumbs of network effects. To not build on open…
Much needed nuance to "gpu indeterminism". Perfectly deterministic training and inference are possible at very tolerable performance cost.
Much needed nuance to "gpu indeterminism". Perfectly deterministic training and inference are possible at very tolerable performance cost.
Minecraft Vibrant Visuals has to be the absolute *worst* implemented shader ever. I remember running shaders looking better than this on my Samsung Galaxy S5 Neo which couldn’t handle YouTube at 1080p60 without lagging and yet PE with shaders ran just fine. Now you’re telling me…
Systems complexity doesn’t just mean deliberately induced complexity. Its often subtle cost that comes from deciding to use off the shelf tools, and losing inspectability for what seemed like a fine trade-off at the time. Every line of code checked into your project is a…
Systems complexity doesn’t just mean deliberately induced complexity. Its often subtle cost that comes from deciding to use off the shelf tools, and losing inspectability for what seemed like a fine trade-off at the time. Every line of code checked into your project is a…

Andrej Karpathy @karpathy
1.4M Followers 1K Following Building @EurekaLabsAI. Previously Director of AI @ Tesla, founding team @ OpenAI, CS231n/PhD @ Stanford. I like to train large deep neural nets.
kache @yacineMTB
196K Followers 6K Following SPONSORED BY FORMLABS - https://t.co/90QFod1lcD - get your 3d printer TODAY prev eng @ x, stripe. yacine_kv on insta I write a subscriber only blog. Subscribe!
Marc Andreessen 🇺�... @pmarca
1.9M Followers 27K Following First name Andreessen, last name Horowitz.
Robert Scoble @Scobleizer
543K Followers 23K Following The best from ML/AI community | Ex-Microsoft, Rackspace, Fast Company | Wrote eight books about the future | Silicon Valley robots, holodecks, BCIs, & startups.
@goth @goth600
70K Followers 9K Following VP, Witchcraft and Propaganda @ 𝕏 | Magic @ 21e8 | “tweets from the void” -redacted
aarya @gd3kr
13K Followers 994 Following three time hackernews survivor & alleged cybercriminal (almost sued by Universal)
nwyin @_nwyin
500 Followers 550 Following
Rohan Choudhury @rchoudhury997
496 Followers 513 Following phd student at cmu https://t.co/pjU847PL2f
aroo @sagenormie
64 Followers 932 Following
Owen Gillett @owengillett
21 Followers 556 Following
manode @_grainierr
245 Followers 4K Following
Connor @utfunderscore
2 Followers 20 Following
Alex Inch @alexinch_ai
121 Followers 835 Following Interested in AI, climate and politics. Just beginning a DPhil on world models @oxfordrobots. Prev MSc @ucl, @tortoise, Physics @UniOfOxford
Samarth Tripathi @samx3499
1 Followers 164 Following
Tal @eiopa
192 Followers 1K Following
major tom @vtomnet
140 Followers 1K Following engineering student @ UC. now: computer-use agents for tetraplegics. dms open.
Hrishbh Dalal @HrishbhDalal
1K Followers 414 Following Machine Learning Lead @ KikiTora Working with AI and LLMs to create an army of Minions. Reach out for projects or colab
Minh Nhat Nguyen @menhguin
11K Followers 6K Following hiring agentic humans @hud_evals / https://t.co/Bz6A6SJeB8 | owned @AIHubCentral (1 million users,acq.) ex climate protester 🦦 don't do the deferred life plan
Seth Karten @sethkarten
1K Followers 593 Following Autonomous Agents | CS PhD @Princeton | Simulation @Waymo | Former @SCSatCMU @Amazon | @NSF GRFP Fellow
KingOfSpadeS @spades_cmd
164 Followers 2K Following Dreamer ◬ | 20 | tech enjoooyer | airship aspirant Building "THE NOTE"
Gio @bug2fix
0 Followers 93 Following
Mohammad Taha Fakhari... @TheSTraveller
533 Followers 4K Following Phd Student @OISTedu. I enjoy to work on the intersection of machine learning, neuroscience, computer science and mathematics. I love football, music and Iran!
Arnie Ramesh @arnie_hacker
4K Followers 3K Following MSc CS @ ETH Zurich | Building stuff @shipfr8 | Belgian, 23
BenIt Pro @BennettBuhner
13K Followers 417 Following 18M | Tech + AI = Life | I talk tech Apple to Samsung | Concept creator, entrepreneur, shit-poster | Building @TheAthenaAI | Wanna work @PrimeIntellect | DM me!
0xdeku @0xdeku
29 Followers 1K Following
GS @SpecialK0025
338 Followers 5K Following Engineer + MBA, Background in Finance, underwriting, risk management with software engineering skills and ability to analyze large volumes of data using ML, AI.
Adi Ganesh @_adiganesh
464 Followers 635 Following Research @openai. Prev. @metaai @nuro @stanford @thielfellowship. Co-created @gradientpub
Josiah @josssiiiah
65 Followers 260 Following Founding Engineer @PlayOasiz | @Stanford Alumni | Improving at Linear Algebra as fast as my brain will let me
Sami @_SamiBG
227 Followers 437 Following Building world models @WayfarerLabs - Prev @BrownUniversity @Bloomberg 🇱🇧/🇺🇸
RobertaJohn @3AZSKr4UVeL30O0
9 Followers 500 Following
Rahul Meena @RMeena73817
2 Followers 72 Following Flutter Developer |2200+points@GFG| C++ Programmer | Python
charlotte @cronjaegerc
129 Followers 1K Following Applied AI @MistralAI // prev. @Meta, @ETH Zurich. Opinions on my own.
Andrej Karpathy @karpathy
1.4M Followers 1K Following Building @EurekaLabsAI. Previously Director of AI @ Tesla, founding team @ OpenAI, CS231n/PhD @ Stanford. I like to train large deep neural nets.
Lex Fridman @lexfridman
4.4M Followers 593 Following Host of Lex Fridman Podcast. Interested in robots and humans.
George Hotz 🌑 @realGeorgeHotz
300K Followers 204 Following President @comma_ai. Founder @__tinygrad__
François Chollet @fchollet
575K Followers 816 Following Co-founder @ndea. Co-founder @arcprize. Creator of Keras and ARC-AGI. Author of 'Deep Learning with Python'.
Sebastian Raschka @rasbt
358K Followers 1K Following ML/AI researcher & former stats professor turned LLM research engineer. Author of "Build a Large Language Model From Scratch" (https://t.co/O8LAAMRzzW).
Yann LeCun @ylecun
954K Followers 765 Following Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.
kache @yacineMTB
196K Followers 6K Following SPONSORED BY FORMLABS - https://t.co/90QFod1lcD - get your 3d printer TODAY prev eng @ x, stripe. yacine_kv on insta I write a subscriber only blog. Subscribe!
elvis @omarsar0
266K Followers 680 Following Building with AI agents @dair_ai • Prev: Meta AI, Galactica LLM, Elastic, PaperswithCode, PhD • I share insights on how to build with AI Agents ↓
AK @_akhaliq
428K Followers 3K Following AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo ,submit papers here: https://t.co/UzmYN5YmrQ
Andrew Ng @AndrewYNg
1.3M Followers 1K Following Co-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
Grant Sanderson @3blue1brown
413K Followers 362 Following Pi creature caretaker. Contact/faq: https://t.co/brZwdQfdif
Jim Fan @DrJimFan
327K Followers 3K Following NVIDIA Director of Robotics & Distinguished Scientist. Co-Lead of GEAR lab. Solving Physical AGI, one motor at a time. Stanford Ph.D. OpenAI's 1st intern.
Emad @EMostaque
291K Followers 24 Following Distributing Intelligence @ii_posts. Founder @StabilityAI.
Jürgen Schmidhuber @SchmidhuberAI
165K Followers 0 Following Invented principles of meta-learning (1987), GANs (1990), Transformers (1991), very deep learning (1991), etc. Our AI is used many billions of times every day.
Google DeepMind @GoogleDeepMind
1.2M Followers 279 Following We’re a team of scientists, engineers, ethicists and more, committed to solving intelligence, to advance science and benefit humanity.
Dwarkesh Patel @dwarkesh_sp
130K Followers 916 Following Host of @dwarkeshpodcast https://t.co/3SXlu7fy6N https://t.co/4DPAxODFYi https://t.co/hQfIWdM1Un
Richard Sutton @RichardSSutton
50K Followers 64 Following Student of mind and nature, libertarian, chess player, cancer survivor. @ Keen, UAlberta, Amii, https://t.co/u8za2Kod54, The Royal Society, Turing Award
Eric W. Tramel @fujikanaeda
2K Followers 734 Following Research Scientist @ Nvidia. Ex: Synth Data @ Gretel & Unlearn, Federated Learning @ Amazon Alexa & Owkin. Postdocs @ INRIA & ENS. Views my own.
Anne Ouyang @anneouyang
7K Followers 926 Following Building @Standard_Kernel, CS PhD student @Stanford | prev: cuDNN @Nvidia, M.Eng, B.S. in CS @MIT | efficient scalable self-improving AI systems | 🌽KernelBench
Standard Kernel Co. @Standard_Kernel
799 Followers 1 Following Building AI Infrastructure with AI; fast kernels go brrr
Ameen Patel @Ameen_ml
1K Followers 1K Following Inference @PrimeIntellect, prev @togethercompute, @AmazonScience, @uwaterloo
sankalp @dejavucoder
17K Followers 599 Following llms and shitposting into crafting ai products and evals dm open to talk on ai engg/post-training/llm stuff
Marco Mascorro @Mascobot
16K Followers 2K Following Partner @a16z (investor in @cursor_ai, @thinkymachines, @bfl_ml, @WaveFormsAI & more) | Roboticist | Cofounder @Fellow_AI | @MIT 35 under 35 | Opinions my own.
davinci @leothecurious
2K Followers 753 Following teaching robots to see by day, learning from nature by night. in search of elegant solutions to the metaproblem. infinitely curious.
Alexander Doria @Dorialexander
19K Followers 4K Following Reasoning models to come. Co-founder @pleiasfr
Infornomics @infornomics
1K Followers 1K Following Data Science, AI & Energy Transition, RSWE economics/ Logic-Weaver, Data-Mancer, Charmer of machines, Wonder-Walker (bestowed by Claude) Goal: emotion spinner
Anemll @anemll
794 Followers 426 Following ANEMLL (pronounced like "animal") Artificial Neural Engine Machine Learning Library, Open Source Project
Eric Zhang @ekzhang1
16K Followers 503 Following Computer systems person, interaction designer. founding eng @modal → dreams of: a simpler, more honest, more human sort of software (people are good, be kind!)
James Bradbury @jekbradbury
13K Followers 9K Following Compute at @AnthropicAI! Previously JAX, TPUs, and LLMs at Google, MetaMind/@SFResearch, @Stanford Linguistics, @Caixin.
Charlie Marsh @charliermarsh
28K Followers 827 Following Building @astral_sh: Ruff, uv, and other high-performance Python tools. Prev: Staff engineer @SpringDiscovery, @KhanAcademy, BSE @PrincetonCS.
gingerBill @TheGingerBill
12K Followers 1K Following I'm a Ginger thus I have no soul. Creator of the Odin Programming Language https://t.co/LWFCLB39eC Working at @JangaFX on EmberGen/LiquiGen/GeoGen/IlluGen
Charles 🎉 Frye @charles_irl
15K Followers 3K Following gpu enjoyer at @modal. he/him. ex @full_stack_dl, @weights_biases (acq. @CoreWeave), phd Berkeley @Redwood_Neuro. try https://t.co/SYWVMCazZ3
Ilya Sutskever's hair... @IlyasHairline
3K Followers 845 Following Follically challenged, but emotionally enriched. On my journey from forehead to backhead. #OnTheMove
Christian Gilli @nirw4nna
217 Followers 233 Following I like building software | Working on a tensor library from scratch: https://t.co/CPhWN3O8rq | Blog: https://t.co/aoUjpbPpw9
Q @qtnx_
17K Followers 549 Following codegen @mistralai (prev. applied research; https://t.co/SDROdHKqTQ), husband
barnii77 @barnii_77
4 Followers 42 Following
You Jiacheng @YouJiacheng
8K Followers 2K Following a big fan of TileLang 关注TileLang喵!关注TileLang谢谢喵! https://t.co/utshC0jrCO 十年老粉
Omar 🍋 @ocornut
18K Followers 157 Following dear imgui https://t.co/iQ5qoqTIRd + test engine / the dragon’s trap @lizardcube, dreams, tearaway, pixeljunk shooter, soul bubbles, meka, smspower
kalomaze @kalomaze
19K Followers 2K Following ML researcher (@primeintellect), speculator • extremely silly jester
Rohan Pandey @khoomeik
39K Followers 2K Following descending cross-entropy to ascend entropy || prev research @OpenAI @CarnegieMellon '23
XWine1 @XWineOne
9K Followers 13 Following Xbox One translation layer for Windows PCs. Not associated with the Wine project. Logo by @Zeealeid
Felix @felix_red_panda
5K Followers 2K Following speech synthesis and LLM nerd, DMs open, working on LLM stuff