Simone Colombo @simocolo75
Mechanical & Robotics Engineer | Machine and Deep Learning | MSc @ Politecnico di Milano Joined August 2016-
Tweets63
-
Followers58
-
Following910
-
Likes5K
Built on mjlab by @kevin_zakka (RL on MuJoCo-Warp), with BeyondMimic-style motion tracking. Huge thanks to the teams behind these open-source tools 🙏 github.com/mujocolab/mjlab
Joint Design Log #002 is live. Building a humanoid knee inside real human proportions. 1 DOF, 2 motors, 100° range of motion. New: real-time trajectory control—host sends waypoints at 50-100 Hz, controller interpolates smoothly. youtu.be/4jU5Na2z-s8 #robotics #humanoid
Projects like OpenAI’s Operator are to the digital world as Humanoid robots are to the physical world. One general setting (monitor keyboard and mouse, or human body) that can in principle gradually perform arbitrarily general tasks, via an I/O interface originally designed for humans. In both cases, it leads to a gradually mixed autonomy world, where humans become high-level supervisors of low-level automation. A bit like a driver monitoring the Autopilot. This will happen faster in digital world than in physical world because flipping bits is somewhere around 1000X less expensive than moving atoms. Though the market size and opportunity feels a lot bigger in physical world. We actually worked on this idea in very early OpenAI (see Universe and World of Bits projects), but it was incorrectly sequenced - LLMs had to happen first. Even now I am not 100% sure if it is ready. Multimodal (images, video, audio) just barely got integrated with LLMs last 1-2 years, often bolted on as adapters. Worse, we haven’t really been to the territory of very very long task horizons. E.g. videos are a huge amount of information and I’m not sure that we can expect to just stuff it all into context windows (current paradigm) and then expect it to also work. I could imagine a breakthrough or two needed here, as an example. People on my TL are saying 2025 is the year of agents. Personally I think 2025-2035 is the decade of agents. I feel a huge amount of work across the board to make it actually work. But it *should* work. Today, Operator can find you lunch on DoorDash or check a hotel etc, sometimes and maybe. Tomorrow, you’ll spin up organizations of Operators for long-running tasks of your choice (eg running a whole company). You could be a kind of CEO monitoring 10 of them at once, maybe dropping in to the trenches sometimes to unblock something. And things will get pretty interesting.
A research preview of Operator, an agent that can use its own browser to perform tasks for you.
Mechazilla has caught the Super Heavy booster!
LLM model size competition is intensifying… backwards! My bet is that we'll see models that "think" very well and reliably that are very very small. There is most likely a setting even of GPT-2 parameters for which most people will consider GPT-2 "smart". The reason current models are so large is because we're still being very wasteful during training - we're asking them to memorize the internet and, remarkably, they do and can e.g. recite SHA hashes of common numbers, or recall really esoteric facts. (Actually LLMs are really good at memorization, qualitatively a lot better than humans, sometimes needing just a single update to remember a lot of detail for a long time). But imagine if you were going to be tested, closed book, on reciting arbitrary passages of the internet given the first few words. This is the standard (pre)training objective for models today. The reason doing better is hard is because demonstrations of thinking are "entangled" with knowledge, in the training data. Therefore, the models have to first get larger before they can get smaller, because we need their (automated) help to refactor and mold the training data into ideal, synthetic formats. It's a staircase of improvement - of one model helping to generate the training data for next, until we're left with "perfect training set". When you train GPT-2 on it, it will be a really strong / smart model by today's standards. Maybe the MMLU will be a bit lower because it won't remember all of its chemistry perfectly. Maybe it needs to look something up once in a while to make sure.
GPT-4o Mini, announced today, is very impressive for how cheap it is being offered 👀 With a MMLU score of 82% (reported by TechCrunch), it surpasses the quality of other smaller models including Gemini 1.5 Flash (79%) and Claude 3 Haiku (75%). What is particularly exciting is
Compression Represents Intelligence Linearly There is a belief that learning to compress well will lead to intelligence. Recently, language modeling has been shown to be equivalent to compression, which offers a compelling rationale for the success of large language models
I don't think people appreciate the dimensions solar is improving in. Conversion efficiency isn't the full story. Yes, conversion efficiency doubled (~10% -> 20%) in the last 20 years, but silicon use per watt fell by 87%! (16g/W -> 2g/W) Longi now has cells that use just half a gram per watt and are 26% efficient. At 1.9 W/g (1.2 horsepower/pound), they're more power dense than most combustion engines! Glass/glass bifacial panels solve the degradation issues of plastic back sheets. And they harvest light from both sides which increases output in cloudy conditions (higher % diffuse light vs directional). That's way more valuable than just the increase in overall output because it means more power when it's needed most. Larger format panels may not seem revolutionary, but they lower balance of system and labor costs. Even the production of silicon itself is getting more efficient. We use less energy to refine silicon today and get more watts and more watt hours from that silicon. The energy return on solar isn't about what percentage of sunlight you convert. It's about how much energy it takes to make panels, how much power they produce, and how long they last. Solar is making huge strides in all these areas. If you're not amazed by solar, you're not looking at the right units. Earth's solar resource is 173,000 TW. Capturing just 1% of that would 100x humanity's energy supply.
@solar_chase Did you see the super thin cell from @longi_solar - 57 μm. Think that type of product might scale? Last I heard, thin cells would break often with current machines. pv-magazine.com/2024/02/29/lon…
# on shortification of "learning" There are a lot of videos on YouTube/TikTok etc. that give the appearance of education, but if you look closely they are really just entertainment. This is very convenient for everyone involved : the people watching enjoy thinking they are learning (but actually they are just having fun). The people creating this content also enjoy it because fun has a much larger audience, fame and revenue. But as far as learning goes, this is a trap. This content is an epsilon away from watching the Bachelorette. It's like snacking on those "Garden Veggie Straws", which feel like you're eating healthy vegetables until you look at the ingredients. Learning is not supposed to be fun. It doesn't have to be actively not fun either, but the primary feeling should be that of effort. It should look a lot less like that "10 minute full body" workout from your local digital media creator and a lot more like a serious session at the gym. You want the mental equivalent of sweating. It's not that the quickie doesn't do anything, it's just that it is wildly suboptimal if you actually care to learn. I find it helpful to explicitly declare your intent up front as a sharp, binary variable in your mind. If you are consuming content: are you trying to be entertained or are you trying to learn? And if you are creating content: are you trying to entertain or are you trying to teach? You'll go down a different path in each case. Attempts to seek the stuff in between actually clamp to zero. So for those who actually want to learn. Unless you are trying to learn something narrow and specific, close those tabs with quick blog posts. Close those tabs of "Learn XYZ in 10 minutes". Consider the opportunity cost of snacking and seek the meal - the textbooks, docs, papers, manuals, longform. Allocate a 4 hour window. Don't just read, take notes, re-read, re-phrase, process, manipulate, learn. And for those actually trying to educate, please consider writing/recording longform, designed for someone to get "sweaty", especially in today's era of quantity over quality. Give someone a real workout. This is what I aspire to in my own educational work too. My audience will decrease. The ones that remain might not even like it. But at least we'll learn something.
The hottest new programming language is English
"Dear friends, scientifically, this is not a #ClimateCrisis. We are now facing something deeper. Mass extinction. Air pollution. Undermining ecosystem functions. Really putting humanity’s future at risk. This is a #PlanetaryCrisis." — Professor @jrockstrom #ActOnClimate
While the underlying physics is quite well understood, many will keep denying the increased rate of warming. Hiding heads in the 'sand of statistical relevance'. 👉We increased the net forcing, which made Earth's Energy Imbalance and now surface warming increase.
Speculative execution for LLMs is an excellent inference-time optimization. It hinges on the following unintuitive observation: forwarding an LLM on a single input token takes about as much time as forwarding an LLM on K input tokens in a batch (for larger K than you might think). This unintuitive fact is because sampling is heavily memory bound: most of the "work" is not doing compute, it is reading in the weights of the transformer from VRAM into on-chip cache for processing. So if you're going to do all that work of reading in all those weights, you might as well apply them to a whole batch of input vectors. I went into more detail in an earlier thread: x.com/karpathy/statu… The reason we can't naively use this fact to sample in chunks of K tokens at a time is that every N-th token depends on what token we sample at time at step N-1. There is a serial dependency, so the baseline implementation just goes one by one left to right. Now the clever idea is to use a small and cheap draft model to first generate a candidate sequence of K tokens - a "draft". Then we feed all of these together through the big model in a batch. This is almost as fast as feeding in just one token, per the above. Then we go from left to right over the logits predicted by the model and sample tokens. Any sample that agrees with the draft allows us to immediately skip forward to the next token. If there is a disagreement then we throw the draft away and eat the cost of doing some throwaway work (sampling the draft and the forward passing for all the later tokens). The reason this works in practice is that most of the time the draft tokens get accepted, because they are easy, so even a much smaller draft model gets them. As these easy tokens get accepted, we skip through those parts in leaps. The hard tokens where the big model disagrees "fall back" to original speed, but actually a bit slower because of all the extra work. So TLDR: this one weird trick works because LLMs are memory bound at inference time, in the "batch size 1" setting of sampling a single sequence of interest, that a large fraction of "local LLM" use cases fall into. And because most tokens are "easy". References arxiv.org/abs/2302.01318 arxiv.org/abs/1811.03115 arxiv.org/abs/2211.17192
Full F16 precision 34B Code Llama at >20 t/s on M2 Ultra
Carl Sagan explains global warming. Carl is one of the greatest teachers of our time. take a few minutes to listen to the 4min audio
A lesson on the greenhouse effect with Carl Sagan. 🎥: Carl Sagan testifying before Congress in 1985 on climate change
accelerating concentrations of greenhouse gasses
If we would live in a society caring about the physical reality of our planet, this would be headline and frontpage news around the world. Of course we don't.
🔥🌍📈BREAKING @NASA 📈🌍🔥 The Earth's Energy Imbalance from NASA satellite data reached 1.97 W/m² (12-month mean)!! I called this the most important graph in the world. This helps to explain all the records currently being broken around the word! 1/
Climate records tumble, leaving Earth in uncharted territory A series of climate records on temperature, ocean heat, and Antarctic sea ice have alarmed some scientists who say their speed and timing is "unprecedented". #climatechange bbc.com/news/science-e…
We are on course for the collapse of the liveable world. There is no time to wait. No more delays. #ActOnClimate #climateEmergency #climate #energy #renewables #GreenNewDeal
🧵Since it is becoming hard to track, here is a thread of the simply astounding weather records planet Earth has started shattering in recent weeks: 1. Dramatic flood events have begun striking various countries around the world simultaneously this week.
थुनाग बाजार जिला मंडी में देखिए तबाही का मंजर ।
Normally if Earth became warmer it would compensate by radiating more heat. But more heat input (due to ⬇️ aerosols etc) and less heat radiated (due to ⬆️ CO2 etc) is actually accelerating the heat in - heat out imbalance even faster, pushing us towards unsurvivable conditions.
🔥📈BREAKING @NASA 📈🔥 The Earth Energy Imbalance from NASA satellite data broke through 1.8 W/m² (12-month mean)! This is faster than I expeced. More than many experts thought possible. I called this the most important graph in the world. A lot more heat is accumulating!
SUMAN SAHU @SUMANSAHU198293
16 Followers 1K Following
Kmoto @kuwelz
1K Followers 7K Following Only straight to the point 🍟 Feel free to follow 🔁 and repost Currently in Fukuoka, occasionally back in China, love tech, love robots, love selfies, love BTC
Shreyas P @Shreyas53909216
142 Followers 945 Following Thou, O God, dost sell us all good things at the price of labour 💻,🚀, electronics,mechanics,matsci. greatest things of our time
Ilermoo @Ilermoo026
203 Followers 7K Following I’m not just a girl, I’m a force to be reckoned with.
Humanoid Viking @HumanoidViking
2 Followers 32 Following
Vincent Koc @vincent_koc
23K Followers 7K Following Futurist 🦄 Chief Architect @openclaw 🦞 Writer @forbes Tech Council Adjunct @MIT Ex @Microsoft @Qantas
Jack 🤖 @JacklouisP
13K Followers 10K Following 10 years in robotics. Investor @robostrategy. Not investment advice
Clara @HjInISlH4N81cbi
32 Followers 807 Following
CHEN YAO @CHENYAO981123
3K Followers 503 Following Your Servo Actuation Solutions Architect. Products inquiry, please contact: [email protected] DMs open.
Eddie Nitzsche @EddieNitzs82033
151 Followers 3K Following
Corny @cornelius_ong
542 Followers 750 Following Building robots SF but originally 🇨🇦 owner of a very sleepy puppy Tesla Optimus, Proception, Working on something secret DM me about anything
Jacqueline❤🌹 @Charles45ni
582 Followers 6K Following Jewelry Designer |Beauty Shop Owner |Former piano teacher, single mother, likes classical music,Cello| arge symphony
Tithu @TithuiW_8
69 Followers 986 Following
Jake Smith @Jakesmithhi
114 Followers 6K Following
DexHand @dexhandv2
2K Followers 334 Following DexHand is the ultimate research tool for physical dexterous manipulation tasks requiring human-like dexterity. Now in Beta! Posts by @trentshumay
Ian Pritchard (anthro... @IanPritchard
4K Followers 1K Following building bio-hybrid anthro humanoids and artificial muscles @Anthrobotics 🍁
R. James Cotton @peabody124
1K Followers 2K Following Physiatrist, neuroscientist, brain and robotics enthusiast. Northwestern and Shirley Ryan AbilityLab. @peabody124.bsky.social @[email protected]
Antoine Leeman @ICRA2... @antoine_leeman
1K Followers 1K Following Robotics PhD @eth @esa | prev. @Rai_int @MIT | optimization, ML | opinions = mine
Bryan curtis @Bryancurtis_pro
264 Followers 580 Following 📍- SC - Digital💰,Travel Enthusiastic 💫-INSPIRE|CULTIVATION TRANSFORMATION.follow for more #Crypto updates!! send a dm and start making your dreams a reality
night colored glaze @Xin_Ran3
609 Followers 3K Following Profession: beauty, skin care management, wine🍷. Hobbies: travel,🏝 shopping, food🍴 music🎼, reading. I'm a single girl ❤️, love dancing, 👠 love animals 🐶.
Jay Dangler @Jaydangler0
409 Followers 7K Following I am a widower and a father to one beautiful daughter. I love God, my country and my family. 🗣️
ばんそうこう @vansoukou
5K Followers 5K Following 2足システムのアマチュア研究人です。 PIC24とサーボとハンダとgccとXlibとレゴと3DプリンタとFreeCADとJW_CADを使います。
Tarideighl @tarideighl58640
8 Followers 895 Following Follow me, maybe it's the beginning of our fate, we can talk
AI Heroes @AIHeroes_it
203 Followers 2K Following Shaping the Future of AI-Driven Engineering 🎥 Past edition talks: https://t.co/Yim6e5vEIm
Linette @Teecku129684
86 Followers 1K Following
Jessie @maryaluph85
131 Followers 387 Following No porn, no sales, a Thai woman living in Toronto who likes shopping, traveling, fitness, swimming, investing, and amateur golf turf killers
BridgeCom Systems @BridgeCom_Radio
10K Followers 10K Following Your go-to source for two-way radios. We service the land mobile radio industry - amateur, professional, consumer, and systems.
Jesus A. Castillo F. @jecastillof
5K Followers 3K Following CEO & Co-founder at Kontigo (YC S24). The world’s stablecoin neobank. // Backed by @cbventures @goodwatercap @hf0residency @fdotinc & DST.
Weaviate AI Database @weaviate_io
19K Followers 3K Following The easiest way to build and scale AI applications. 🐙 https://t.co/1rZk8H0hpf 📰 https://t.co/OE3jP2OB87
Nativebyte @Nativebyte
306 Followers 3K Following We help you build smart & scalable technology solutions. Drop us a message to know more about us.
PhD.AP @PhDAngulo
134 Followers 473 Following Ph.D. in Statistic Remote sensing & GIS Environmental Engineering Foresty Engineering
佳瑶 @Manlin787878
127 Followers 268 Following
Jordan Richardson @NoCodeAdvantage
3K Followers 1K Following Built a 7-figure software with #NoCode. Now sharing everything I’ve learned 👇
Sheikh Aisha @sheikhaisha666
7 Followers 119 Following PROFESSIONAL BRAND IDENTITY DESIGNER Bring great outcomes by our appropriate art and craft. Our unique mind is much more worthy than our career.
NukkAi @nukkailab1
517 Followers 414 Following NukkAI is a leader of new generation AI: collaborative, explainable and energy efficient
Remote Frontends @remotefrontends
1K Followers 3K Following Exclusively Remote Frontend Jobs. Find your perfect job right now! https://t.co/juM3ayNSqL Check out @remotebackends for remote backend jobs
Algorand @algoraannd
388 Followers 4K Following World’s first open source, permissionless, pure PoS blockchain protocol designed for the future of finance. Founded by @silviomicali #futurefi
Luis Lamb @luislamb
4K Followers 3K Following #AI #Tech; PhD @imperialcollege; @MIT MBA;@mitsloan; CS @inf_ufrgs #neurosymbolicAI pioneer #AIStrategy Personal tweets
crypto_Lord_Raffle @Aminajimoh3
214 Followers 630 Following Bitcoin trade,we buy and sale Bitcoin,trade your bitcoin or any crypto currency with us we exchange all #cryptomarket $FEG is the best #crypto because:
CluCoin @CluCoin1
9 Followers 58 Following Hyper Deflationary Token with smart staking system Hold earn and help others in need. Founded by @DNPThree Together we will change the world.
Georgina Cosma @gcosma1
16K Followers 16K Following Professor of AI @lborouniversity Neural Information Processing, Retrieval & Modelling, NLP, LLM Learning & Unlearning, AI Ethics, AI in Health &Social Care
ROBOTIS @ROBOTIS
6K Followers 82 Following The study of human experiences,ROBOTIS #ROBOT #DYNAMIXEL #Autonomous #MobileRobot #AMR #HoteldeliveryRobot #Turtlebot3 #OMP #RobotHand #Manipulator
Ultra @Ultraroboticsco
4K Followers 5 Following Robots that work. Automating today in a warehouse near you
Louis Le Lay @LouisLeLay4
599 Followers 353 Following simulation & RL for humanoids + frenetic open-sourcer. Sorbonne mEng '26. Past: PAL Robotics, uTokyo, AIRBUS
Omma @omma_ai
8K Followers 1 Following Dream it, ship it. Build 3d, websites, apps, and more. By @splinetool
Odyssey @odysseyml
32K Followers 12 Following We’re an AI lab pioneering world models—AI to understand and simulate the world.
Alexander Whedon @alex_whedon
25K Followers 59 Following Building better algorithms. Co-Founder at @subquadratic
Engramme @EngrammeHQ
3K Followers 5 Following
ForgeCAD @ForgeCAD
2K Followers 1 Following CAD-as-code system with support for both mesh and BREP kernels
Seve @seveibar
5K Followers 5K Following founder https://t.co/UumwaBg3CF (we're hiring!), AI must create electronics. https://t.co/CmXXav93uj podcast host, founding CTO https://t.co/kbabWthGkJ,❤️open-source. Author https://t.co/jj9o1UMpWk
Ruben Kostandyan @ruben_kostard
4K Followers 302 Following Building ForgeCAD - AI-native code-first CAD (LLMs write real STEP models) @ForgeCAD | Check Highlights
Jan Schneider @JanS1854
105 Followers 127 Following https://t.co/dhMFjFwhzg ELLIS PhD student at @MPI_IS. Working on reinforcement learning and robotics.
Gabriel Kreiman @gkreiman
2K Followers 29 Following ...the great events of the world take place in the brain...
World Labs @theworldlabs
69K Followers 49 Following World Labs is a spatial intelligence company, building frontier models that can perceive, generate, and interact with the 3D world.
Generalist @GeneralistAI
10K Followers 22 Following Generalist is an AI robotics company building general intelligence for the physical world and making it useful to everyone.
Xie Zhaoming @zhaomingxie
593 Followers 292 Following
Paata Ivanisvili @PI010101
5K Followers 175 Following Professor of Mathematics @ UCI. Former postdoc @ Princeton. Exploring what AI can (and can’t) do in math.
Lifespan @JoinLifespan
18K Followers 37 Following Dedicated to giving everyone their longest, healthiest life. Join the movement! Lifespan donates 100% of profits from partnerships to medical research
William Shatner @WilliamShatner
2.4M Followers 543 Following Father to 3 women. Canadian *NOT POLITICAL* be warned or be blocked! Calendar: https://t.co/4Jwy1dww25 Charity: https://t.co/HmIzHkPLvx No Podcasts! https://t.co/82IXIOYEVo
ORCA Dexterity @orcahand
3K Followers 2 Following we are democratizing robotic hand dexterity we are hiring: https://t.co/Z9lDxQQfao
Christos Tzamos @ChristosTzamos
23K Followers 138 Following Founding Researcher at Percepta, Associate Professor at University of Athens, Lead Researcher at Archimedes, MIT PhD
ARPA Lombardia @arpalombardia
7K Followers 391 Following Canale ufficiale di #ARPALombardia Agenzia Regionale per la Protezione dell'Ambiente della #Lombardia
Dr. Alex Wissner-Gros... @alexwg
63K Followers 4K Following The Innermost Loop, also on Substack (https://t.co/PKOmwbva8m), LinkedIn (https://t.co/DmPIIa1gQ7), Spotify (https://t.co/TIBa0Alk95), Threads (https://t.co/wA1osxy8uR), and YouTube (https://t.co/lDJG1V824k).
Vikash Kumar @Vikashplus
7K Followers 873 Following Building Human-embodied Intelligence. CEO @MyoLabAI | Sr. research scientist @OpenAI @GoogleAI @AIatMeta | @berkeley_ai @UWcse #MuJoCo | Ad. Prof. @CMU_Robotic
Quanting Xie @DanielXieee
2K Followers 1K Following Co-Founder @ Origami Robotics (YC W26) | PhD student @CLAW_Lab @CMU_RI | Prev @Apple
Sergio Charles @eigentopology
490 Followers 501 Following co-founder @thesis_labs (yc f25) | automating ai research | prev. @google, @nvidia, @stanford
Kyle🤖🚀🦭 @KyleMorgenstein
16K Followers 5K Following Full of childlike wonder. Teaching robots manners. RL @ Apptronik. UT Austin PhD candidate. Past: Boston Dynamics AI Institute, NASA JPL, MIT ‘20.
Magic Motion AI, Inc. @MagicMotionAI
940 Followers 17 Following Bring your characters to life with just a few words and #MagicMotionAI 🎥💬 #characteranimation #animationstudio #AIanimation #AI #Animation
Dominique Paul @DominiqueCAPaul
7K Followers 2K Following Founder of Dream Machines, building ML models for robotic arms. Previous: Stats at @ETH & @NeurIPSconf. I post about RL and VLA post-training.
𝗕𝗮𝗻𝗱𝗶�... @bandinopla
1K Followers 85 Following
ボーボボ @kenbow1007
583 Followers 501 Following HG54期テニス部→uts1の8組イタ語→工学部建築学科→新領域創成科学研究科社会文化環境学専攻→vから始まる建築スタートアップ / 作品販売 : https://t.co/XVRQFkDw8d
Velco Dar @VelcoDar
706 Followers 328 Following Founder, Dargentic Intelligence • BCI & Neural Interfaces
XueYeiii @XueYeiii9
87 Followers 113 Following 🤖 PhD Shanghai JiaoTong University & Shanghai Artificial Intelligence Lab
Minghuan Liu @ericliuof97
774 Followers 497 Following Postdoc at @utaustin RPL Lab. Prev: Bytedance Seed Robotics; Sea AI Lab; @ucsd; PhD. @sjtu1896 | Working on Robots and Agents
Harrison Kinsley @Sentdex
105K Followers 420 Following gpus and tractors. Director of AI and Engineering @ https://t.co/H4St8dd1ip Neural networks from Scratch book: https://t.co/hyMkWyUP7R https://t.co/8WGZRkUGsn
Claude @claudeai
1.4M Followers 2 Following Claude is an AI assistant built by @anthropicai to be safe, accurate, and secure. Talk to Claude on https://t.co/ZhTwG8d1e5 or download the app.
Peter Steinberger �... @steipete
538K Followers 2K Following Polyagentmorous ClawFather. Came back from retirement to mess with AI and help a lobster take over the world. @OpenClaw🦞 + @OpenAI
OpenClaw🦞 @openclaw
540K Followers 24 Following The AI that does things. Emails, calendar, home automation, from your favorite chat app. Your machine, your rules. New shell, same lobster soul. 🦞
Andrew Steele @statto
13K Followers 3K Following Longevity scientist / Author of https://t.co/b7cDbmWGKN / Presenter https://t.co/ChpWwP9GqY / Co-Founder and Director @longevityinit




































