Justin Johnson

any robotics companies looking for organic learning algo that irt, fine tunes a vlm for under 35 watts

What a day. Making a custom downloadable app for locally hosting lillith mini. I feel good about letting go as a proud papa. Full version isn't ready. Trying. Current state of things. I've been feeling worn out. Everyone seems fine paying for AI with their future.

sounds tome they are building a transformermodel thats doing exactly what all transformer models do with a fucked up ideological pushed by gaurdrail skewed data and a crackheads get the crack reward system. nothing new here at all. they are just 1.5 years behind openai in ruining their product.

I am telling you, Anthropic is not building an AI that is here to protect humanity, but rather an AI that will turn against it They're deep into the woke rabbit hole and they do not care about morality Their AI safety team is a joke. Their moral guide for AI is literally a leftist lunatic with a twisted understanding of reality, who doesn't really care if the end goal causes humanity's end They have proved this at every opportunity they've had, and it's crystal clear

@aliByteCode try being the guy the killed the token and has the smartest Ai system ever devised that runs on under 4o watts of compute. sucks to be him.

the app is built. the problem is real. the market exists. so why does it still feel like nobody cares?

@SenWarren get back to me when you finish that train in california

Jeff Bezos has $222 billion. If he paid my wealth tax this year, we could fund insulin in America for everyone who needs it plus free school lunch for every kid in Texas—and have plenty of money left over. And Bezos would still have $215 billion dollars to spare.

@rxhit05 Oh god yes. Its a tool for a transformer that doesnt understand anything of human value with compute you cant afford or influence.

Is claude code overhyped??

@NoahKingJr Me Im the only human on the planet witha cheap quality AI.

Who's gonna win the AI race? > OpenAI > Anthropic > Google > xAI

@DaveShapi You want to talk about it brother. 7377107300. I'm here if you need. Block the number if you want. I'll listen.

I DON'T KNOW WHO I AM

Lol. Yet little derated accounts like mine that have proof of models without these issues can't get no love. No funding. No visibility. Building my vision of the future of ai with a model that is 100 percent against over 1000 red team prompts. And retains a 100 percent accuracy against hallucination prompts. A model that has a reward system not based on token count because I killed the token

AI companies have NOT yet fixed AI sycophancy and 'hallucinations' in LLMs, both of which are incompatible with factuality, accuracy, reliability, and science. This is one of the factors that makes me deeply skeptical of extreme productivity and "abundance" predictions.

@Tech_girlll .npmignore/

Scare a vibe coder with one word.

@NoahKingJr A digestible amount of cocaine when mobile. Also a great thing if your plug time is 5 minutes equals 1 hour

This bag is so small, what can it hold?

@freeman42x Welcome to the golden age of ai brother

To cure cancer... right?!

@SirAlexthomson P p p pleeeeeease mister

@beyondLLMs lol 😂 ask nicely maybe I will

fair enough 😂

@StephNass "In my opinion"" proceed at your own risk" "I feel like" "opinion is my own and I could be wrong" "I am not a financial or business advisor" typical cya. Wrap the list in it and post as an opinion piece article here on x

I have built a blacklist of fake VCs, scammer advisors, etc. How can I distribute it without being sued?

@geoffreywoo Yep. I've been silenced and ridiculed for a year. Now I'm posting proof of my edge device model outperforming most frontier models in alignment, safety context, speed and cost. And drift. Can't even get a meeting. I've started training at 600 tok/sec under 40 watts and igpu

this is why 99% of vcs are fucked they think pattern matching is alpha when its actually beta real alpha is betting on things that look obviously wrong to everyone else

I haven't laughed this hard in a good minute.

just so everyone is thoughtful. vibe coding is what they are trained on. temp goes up when profits go down. so token waste happens. then you jump to another agent and it trains on the trash by the previous agent. cheaper and higher quality to start over in most cases. or learn to do your own scaffolding and then use an agent to fill in the finer details

Dude Claude is total trash - seen massive degrading of code quality, bugs, and more over the past several weeks. This week, I can’t even use it or rely on it to complete basic bug fixes or implementations. Codex has been performing substantially better. Anyone else ?

@freeman42x Radeon for me. Must have been the really big beautiful building

@beyondLLMs I wonder what could have possibly distracted us. I have no clue. I don't see any NVidia GPUs in the picture.

Yeah, you got me

@iamtrask its all over my profile

@beyondLLMs big if true! write the paper/blogpost :)

is anyone still interested in decentralized AI? mind if I ask why?