bl1ss (@SteveDJ_) header
bl1ss (@SteveDJ_) profile picture

hey | llama | hedgehog

Joined August 2009
Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)Photo by bl1ss (@SteveDJ_)
bl1ss retweeted
The new tax in the Netherlands is trending worldwide just not in the country that created it
96
113
10
2,094
78,384
HahahaahahahahahahaahahhaaHahahaahahahahahahaahahhaaHahahaahahahahahahaahahhaa
Sinds de dood van de 17-jarige Lisa uit Abcoude hebben meerdere gemeenten maatregelen genomen om de veiligheid van vrouwen op straat te vergroten. Ook in Amsterdam zijn onlangs nieuwe plannen aangekondigd om de openbare ruimte veiliger te maken. Wat gaat er concreet veranderen? Wethouder openbare ruimte in Amsterdam Melanie van der Horst: #pauwendewit
7
160
Wel leuk dat de president-directeur van de Nederlandse Spoorwegen de zaak daar zo goed op de rit heeft, je hoort er eigenlijk nooit klachten over, dat hij tijd over heeft om de kabinetsformatie ook nog even op gang te helpen.
484
485
48
4,009
134,957
Why is there no mass-adoption? lol
you probably need an optimal mix of klingon, ferengi and vulcan for bitcoin. klingon - mad about spam, sending social message. ferengi - to get that economics drive everything. vulcan - to do the game-theory and coding. the federation is multi-species, but works pretty well.
2
3
141
Things you love to see <3 @boldleonidas good for you after shorting the bottom for the rest of us
1
3
104
Respect the pump
1
2
159
I'm about to be up @BiglyGG
1
35
A society that burns neighbourhoods and calls for the defunding of the police for the guy on the left and largely ignores the public butchering of the woman on the right is one that is drowning in the infinity pool of Suicidal Empathy.
2,589
20,428
1,325
112,179
37,839,908
bl1ss retweeted
Explanation of the current npm hack In any website that uses this hacked dependency, it gives a chance to the hacker to inject malicious code, so for example when you click a "swap" button on a website, the code might replace the tx sent to your wallet with a tx sending money to hacker But in your wallet you'd still see the bad tx and need to approve it, its not like you'll instantly get drained Furthermore, this will only impact websites that pushed an update since the hacked npm package was published, as other projects will have the old version And most projects pin their dependencies, so even if they push an update they'll keep using the old safe code So your wallet is safe and the effective impact area is much smaller than "all websites", but since you cannot really know if a project pinned dependencies, or if they have some dynamically downloaded dependency (very unlikely), it's just safer to avoid using crypto websites till this blows over and they clean up the bad packages
305
812
164
3,829
635,477
life is incredibly easy all you need to do is sleep 8 hours, drink enough water, lift weights, do yoga, stretch daily, walk outside, get morning sunlight, eat whole foods, cut processed sugar, avoid seed oils, eat more protein, track calories, supplement wisely, take magnesium, take vitamin d, get regular bloodwork, sauna, cold plunge, deep breathing, meditate, journal, gratitude practice, spend time in nature, reduce alcohol, quit smoking, reduce porn, limit social media, read books, learn new skills, write daily, practice a language, play an instrument, create art, travel, meet new people, strengthen old friendships, call family, work on relationship, flirt with partner, have regular sex, laugh often, join a community, volunteer, help others, set goals, break goals into steps, plan the week, review progress, block distractions, delete useless apps, focus on one task at a time, use a calendar, build routines, stick to routines, keep promises to self, build financial literacy, budget monthly, track expenses, pay off debt, save money, invest money, learn about crypto, buy btc, diversify investments, hold some gold, own real estate, build side income, negotiate pay, build personal brand, write online, share knowledge, start newsletter, build network, attend events, find mentors, mentor others, practice public speaking, learn marketing, learn sales, improve copywriting, practice storytelling, build leverage, automate tasks, outsource low value work, prioritize health, prioritize wealth, prioritize love, optimize environment, declutter home, clean room daily, fix sleep environment, use blackout curtains, wear blue blockers, avoid screens before bed, take naps when needed, practice mobility, do grip work, strengthen feet, fix posture, train balance, train coordination, do sports, dance, swim, climb, run sprints, train martial arts, condition tendons, learn survival skills, cook at home, try fasting, try time-restricted eating, experiment with diet, learn breathwork, train vagus nerve, practice kegel exercises, optimize hormones, get medical checkups, track hr and sleep, track body fat, track strength, track mood, set challenges, embrace discomfort, build discipline, face fears, say no more, stop gossip, stop complaining, forgive, let go of resentment, focus on solutions, stay curious, stay humble, learn history, study philosophy, study economics, study psychology, study human nature, keep a dream journal, analyze dreams, pursue aesthetics, design personal style, upgrade wardrobe, groom well, smile more, improve posture, make eye contact, move with purpose, speak clearly, articulate better, listen more, ask better questions, practice charisma, cultivate presence, enjoy silence, embrace boredom, limit cheap dopamine, replace with earned dopamine, create beauty, seek beauty, collect memories not things, capture photos, write stories, explore new places, sit in coffee shops, observe people, think deeply, reflect often, refine principles, live by principles, seek mastery not mediocrity, embrace excellence, respect time, respect energy, protect attention, protect health, build resilience, adapt quickly, recover fast, stay consistent, keep momentum, think long-term, live simply, enjoy luxuries selectively, love fully, stay strong, stay free.
125
160
26
1,522
123,155
Cute, Pikachu won the new name round! 🥳 Our 86th Community Adoption. 🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔🦔
1
2
6
105
𝗔𝗻𝗰𝗶𝗲𝗻𝘁 𝗚𝗿𝗲𝗲𝗰𝗲 𝗵𝗮𝗱 𝘁𝗵𝗶𝘀 𝗰𝗿𝗮𝘇𝘆 𝗹𝗮𝘄: You had to get high before you could rule. And no ruler was exempt. Marcus Aurelius, Plato, Cicero, Augustus. All drank the same ego-dissolving potion. They called it: The Eleusinian Mysteries. The drug cult that birthed democracy 🧵
78
219
53
1,481
152,850
If we just ignore the problem maybe it'll go away on it's own? Right? Right..?
NEW - All parties except the AfD have pledged to speak "only positively about migration" in the upcoming local election campaign in the West German city of Cologne.
1
30
Which one of you is this? lmao
15
Let's go silver. Breaking out
14
Oof
Replying to @Polkadot
but you never did anything
21
Nah bitch
BREAKING: Kanye claims his Instagram was hacked to promote a fake crypto coin.
93
16
1
935
104,947
Let me paint the picture. Say you want a Humanoid to cook a delicious dinner while you sit on the couch and watch Netflix. How would this work at a technical level since the Humanoid requires several brains to make your dinner? The robot’s intelligence isn’t monolithic, but a team of AI modules combining slow deliberation with fast reflexes (System 2 + System 1 design). Its vision-language-action (VLA) model splits cognition into a reasoning module and a reactive control policy. As the robot is running on a multi brain cognitive architecture, it would spin up a dedicated “chef” operator to handle your request, such as surveying the kitchen with its cameras, looking up a recipe, then directing its limbs to start chopping vegetables. These brains can be broken down into the following operators. Brain #1: To cook a delicious dinner, you need an executive planner. It interprets your command (“prepare dinner”) to determine the goal (make pasta). Using natural language understanding, it decides what sub tasks are needed (find ingredients, cook pasta, set table etc) and which other brains should handle each. It coordinates the multi agent system: activating specialized modules for vision, knowledge, and motion. This deliberative brain (system 2) makes top level decisions, lays out the approach, and allocates responsibilities before any physical movements start. Brain #2: Now that you have the recipe, you’ll need some robot eyes and spatial awareness. It processes camera feeds to identify ingredients, tools, and their locations in the kitchen. Using advanced computer vision, it sees the cutting board, the veggies in the fridge, the knife on the counter etc. It builds a 3D map of the environment and tracks relevant objects (like where the salt or pans are). This perceptual brain (System 2) runs slower than reflexes, but provides accurate scene context for planning. By recognizing all the pieces involved, it educates the robot in the real world. Brain #3: This brain acts as the robot’s knowledge base and memory (System 2). It retrieves and parses information needed for the task, in this case, a suitable recipe and cooking instructions. It might query an online cookbook or its internal database for a pasta recipe, then interpret the steps (boil water, chop garlic, etc). It recalls facts about the kitchen (like where spices are kept) and past cooking experiences. Essentially, providing semantic understanding and world knowledge. Then computes abstract instructions (caramelize the onions) into concrete parameters (temperature, timing) that the robot can execute, ensuring the plan aligns with your preferences. Brain #4: With the goal and environment clarified, we’ve devised a detailed game plan. It breaks down the high level goal into ordered actions and conditional steps. It schedules tasks (sometimes in parallel, like preheat the oven while chopping vegetables) and sets milestones (water boiled, sauce ready). It also tracks progress and can re-plan on the fly if something changes (say an ingredient is missing). It then hands off this action sequence to the motion level brains for execution. Another System 2 brain. Brain #5: Time to move from the System 2 architecture and move over to System 1, translating the plan into concrete robot movements. For each action (like “walk to the fridge” or “chop carrots”), it generates applicable trajectories for the robot’s body and limbs. This module handles path planning and inverse kinematics, calculating joint paths and angles so the robot moves smoothly without collisions. It typically applies learned motor policies (like a diffusion transformer policy) to produce fluid motions for complex tasks. If Brain 4 says to retrieve a pot from the fridge, Brain 5 figures out how to get the robot there and how to grasp the pot. Where it coordinates multiple limbs when needed (using two hands to lift a heavy pot for example). High level intent turns into a convergence of hardware and software moving in motion Brain #6: Once a motion plan is set, it’s time to execute. This low level System 1 control brain drives the robot’s actuators (motors and joints). It continuously reads sensors (joint angles, force, balance) and sends control signals to follow the trajectory. Using control loops (PID controllers, model predictive control etc) to keep precision, if the robot starts to tip or a knife veers off course, it corrects instantly. These are the reflexes and fine motor skills operating at millisecond speeds. As the robot slices a carrot, Brain 6 modulates the force and adjusts the blade angle to get uniform slices without slipping. It’s like the subconscious “muscle memory” of the system, handling low level details automatically. Brain #7: The final piece is focusing on continuous improvement. During and after the dinner prep, it analyzes performance. Did it spill anything? Was it too slow at stirring? This module uses reinforcement learning and self calibration to update the robot’s models over time. The robot’s core skills were initially trained on massive human demonstrations and trial and error, but you need to be continuously fine tuning them. If it discovers a more efficient dicing technique or a better spatula grip, it updates its policy so that the next dinner goes even smoother. This adaptive brain allows the humanoid to become more skilled with experience. Codec: Operators in Action How does Codec’s architecture tie these brains together? Each “brain” runs as a separate Operator module in the robot’s AI system. Codec’s Fabric orchestration provides each operator with its own secure, sandboxed environment. Meaning, the vision module, language/logic module, planning module etc, all run in isolation yet communicate through defined interfaces. If one module crashes or has errors, it won’t bring down the whole robot, the others keep running safely. This modular design also makes it easy to update or swap out one brain without affecting the rest, and to add new specialized operators as needed. This operator approach directly supports the multi brain framework. When you request dinner, the robot’s executive brain (Brain 1) can spin up a “chef” operator dedicated to that task, while other operators handle perception and control in parallel. Each operator only has access to the resources it needs (for example the recipe agent might have internet access to fetch instructions, whereas the control agent only interfaces with hardware), which improves safety. Codec’s modular, sandboxed design is the glue to all these diverse skills working together, similar to microservices in software, enabling the humanoid to reliably handle complex tasks like cooking dinner from scratch. This is why $CODEC will be the primary infra for Robotics.
You’ll see foundation models for Humanoids continually using a System 2 + System 1 style architecture which is actually inspired by human cognition. Most vision-language-action (VLA) models today are built as centralized multimodal systems that handle perception, language, and action within a single network. Codec’s infrastructure is perfect for this as it treats each Operator as a sandboxed module. Meaning you can spin up multiple Operators in parallel, each running its own model or task, while keeping them encapsulated and coordinated through the same architecture. Robots and Humanoids in general typically have multiple brains, where one Operator might handle vision processing, another handling balance, another doing high level planning etc, which can all be coordinated through Codec’s system. Nvidia’s foundation model Issac GR00T N1 uses the two module System 2 + System 1 architecture. System 2 is a vision-language model (a version of PaLM or similar, multimodal) that observes the world through the robot’s cameras and listens to instructions, then makes a high level plan. System 1 is a diffusion transformer policy that takes that plan and turns it into continuous motions in real time. You can think of System 2 as the deliberative brain and System 1 as the instinctual body controller. System 2 might output something like “move to the red cup, grasp it, then place it on the shelf,” and System 1 will generate the detailed joint trajectories for the legs and arms to execute each step smoothly. System 1 was trained on tons of trajectory data (including human teleoperated demos and physics simulated data) to master fine motions, while System 2 was built on a transformer with internet pretraining (for semantic understanding). This separation of reasoning vs. acting is very powerful for NVIDIA. It means GR00T can handle long horizon tasks that require planning (thanks to System 2) and also react instantly to perturbations (thanks to System 1). If a robot is carrying a tray and someone nudges the tray, System 1 can correct the balance immediately rather than waiting for the slower System 2 to notice. GR00T N1 was one of the first openly available robotics foundation models, and it quickly gained traction. Out of the box, it demonstrated skill across many tasks in simulation, it could grasp and move objects with one hand or two, hand items between its hands, and perform multi step chores without any task specific programming. Because it wasn’t tied to a single embodiment, developers showed it working on different robots with minimal adjustments. This is also true for Helix (Figure’s foundation model) which uses this type of architecture. Helix allows for two robots or multiple skills to operate, Codec could enable a multi agent brain by running several Operators that share information. This “isolated pod” design means each component can be specialized (just like System 1 vs System 2) and even developed by different teams, yet they can work together. It’s a one of a kind approach in the sense that Codec is building the deep software stack to support this modular, distributed intelligence, whereas most others only focus on the AI model itself. Codec also leverages large pre trained models. If you’re building a robot application on it, you might plug in an OpenVLA or a Pi Zero foundation model as part of your Operator. Codec provides the connectors, easy access to camera feeds or robot APIs, so you don’t have to write the low level code to get images from a robot’s camera or to send velocity commands to its motors. It’s all abstracted behind a high level SDK. One of the reasons I’m so bullish on Codec is exactly what I outlined above. They’re not chasing narratives, the architecture is built to be the glue between foundation models, and it frictionlessly supports multi brain systems, which is critical for humanoid complexity. Because we’re so early in this trend, it’s worth studying the designs of industry leaders and understanding why they work. Robotics is hard to grasp given the layers across hardware and software, but once you learn to break each section down piece by piece, it becomes far easier to digest. It might feel like a waste of time now, but this is the same method that gave me a head start during AI szn and why I was early on so many projects. Become disciplined and learn which components can co exist and which components don’t scale. It’ll pay dividends over the coming months. Deca Trillions ( $CODEC ) coded.
14
17
1
81
9,859