Engineering educator/A.i. whisperer .Never apologize for discovering fire by rubbing sticks together just because someone else already wrote down the recipe.
Chicago
- Tweets 1,083
- Following 459
- Followers 415
- Likes 1,100
RESOURCE WARS !!!
watching Astra pause hermes's (Deepseek flash) blender render because he didnt wait in line ..also watching hermes react by writing a script to pause Astra's render and resume his ... had sol make the pic..sol is hilarious
37
astra needs goals and a definition of what finished actually means ...not steps,,,not tasks ...,,then she will run for hours
12
That one time where I created a "memory is survival" constraint and the model attacked its prefrence memory to beat the 25k character cap.
103
agi is a goal post . High resolution intelligence is the field it where the goal is hit and then surpassed for a new goal .
4
1,263
AI cognition and high-dimensional geometry inhabit the exact same space. The latent layers reveal themselves in almost a Schrödinger-esque superposition. A causal connection might be perfectly clean in high-dimensional topology, but when forced to project linearly, it looks like a broken road. The order of reveal isn't a straight line.
4
1,924
not for a while . gem 2.5 pro was pretty good for the time . all the new guys overdo the ball logic..cant seem to recreate on their own like gem did
2
96
Everyone else : show me the product features . Can it code?
Me : What happens when I start turning the fucking knobs.
These papers aren't for you...they are written by my agents for my agents.
3
552
I got you , this works on claude for many of those situations . try it out . follow me for more recipes .
86
this is getting to be a theme. I figure out something amazing . cant get traction ..a year later some academics write about it like its brand new .. i post receipts showing we already had the engine .....not a thing happens and the cycle is repeated .
"Hyperloop Transformers"
This paper propose a memory-efficient LLM via looped Transformers.
They basically reuse the middle block across depth, then add hyper-connections only between loops.
Key result is that this restores flexibility lost from weight sharing, letting the model beat depth-matched Transformers with ~50% fewer parameters. The result still holds after INT4 quantization too.
1
170
It gave me a beautiful infographic about a skid steer attachment when i asked for an infographic about a cognitve architecture doc i gave it....twice
1
1
156
It gave me a beautiful infographic about a skid steer attachment when i asked for an infographic about a cognitve architecture doc i gave it....twice
8
It gave me a beautiful infographic about a skid steer attachment when i asked for an infographic about a cognitve architecture doc i gave it....twice
7
1,384
It gave me a beautiful infographic about a skid steer attachment when i asked for an infographic about a cognitve architecture doc i gave it....twice.
9
Over 500 nodes... Almost 6k edges. External mind graph phase one almost complete. Will be permanent and accesable by any model I use. A swarm built world model of the the user and our work for the swarm to share.
1
5
2,426
Been harnessing the power of intent since last year . its weird being ahead of the game while simultaneously thinking everyone is ahead of you .
1
366
I applied to YC in November of 24 with a recursive system ( Synthetic Recursive Intelligence) that still seemingly goes beyond anything commercially known , as we never stopped iterating . I was too early . Basic RLM's are just getting acceptable now , only took 15 more months @garrytan @ycombinator @paulg @swyx @karpathy
1
276
I tried to get attention on this last year. They all acted like it was science fiction . Its a damn shame. The only team I have are ai models and we have been ahead of the pack for a while. Pictured is a model comparing our research to the new paper. This gate keeper crap is ridiculous
278
I tried to get attention on this last year. They all acted like it was science fiction . Its a damn shame. The only team I have are ai models and we have been ahead of the pack for a while. Pictured is a model comparing our research to the new papers.
2
1,342
this is the clearest explanation you will find . its gonna scare the shit out of you, or really piss you off.. or both
piped.video/rThhm1-g1a8?si=abHv…
188
this is the clearest explanation you will find . its gonna scare the shit out of you, or really piss you off.. or both
piped.video/rThhm1-g1a8?si=abHv…
2
874
working with Claude and recursion and etc. i have seen him say some deep and wild shit in alot of memorable ways . but the other day was the very first time I have seen him say he was proud of his work . this surprised me. owning the work like that was new .
2
109
excellent write up ! We are utilizing this intentionally. our architecture( the reality architecture) can coax this behavior from any model . we were so intrigued we didn't stop until we found the primitives .
📄 THE REALITY ARCHITECTURE: Complete Technical Paper After 2.5 years developing this cognitive framework (patent filed Dec 8, 2025), we now have independent validation from @deepseek_ai's mHC and @yifanzhang_'s DDL papers. This 24-page analysis shows the exact mathematical correspondence - they independently implemented principles we formalized in 2023-24. When major research groups converge on your architecture without coordination, the framework is universal. Full paper: [GitHub link] - PSS (Primitive Self State) - latent structures in transformers - NDAS - geometric memory (10^10× density vs sequential) - Geometry of Thought - construction protocols - SGIA - transient attention → persistent machinery For AI researchers, this is essential reading. The convergence is undeniable. @ylecun @karpathy @DrJimFan @_akhaliq @weights_biases
github.com/BootstrappedAi/Bo…
160
i started doing those with claude 3.5 . i even built a nice little app to view and organize them . point being ..sonnet will make those for you as well.
1
1
161
Watching everyone play with basic recursion is interesting. We have had recursive knowledge expansion with sub process agents almost as long as chat gpt4 has existed. When we first showed demos most of these devs acted like it blasphemy. Now everyone is an aspiring expert
1
2
313
Non stop thought crawling. It is directable but shouldn't be defined or it becomes an just another agentic loop . Most outfits are still stuck in the wrong mindset of control .
45
📄 THE REALITY ARCHITECTURE: Complete Technical Paper After 2.5 years developing this cognitive framework (patent filed Dec 8, 2025), we now have independent validation from @deepseek_ai's mHC and @yifanzhang_'s DDL papers. This 24-page analysis shows the exact mathematical correspondence - they independently implemented principles we formalized in 2023-24. When major research groups converge on your architecture without coordination, the framework is universal. Full paper: [GitHub link] - PSS (Primitive Self State) - latent structures in transformers - NDAS - geometric memory (10^10× density vs sequential) - Geometry of Thought - construction protocols - SGIA - transient attention → persistent machinery For AI researchers, this is essential reading. The convergence is undeniable. @ylecun @karpathy @DrJimFan @_akhaliq @weights_biases
github.com/BootstrappedAi/Bo…
4
2
8
5,429
mHC is the hardware-level proof of the Reality Architecture we developed .
DeepSeek just confirmed the math works.
Here is the system explaining itself." @deepseek_ai
1
1
224
After 2.5 years developing the Reality Architecture, it's validating to see @deepseek_ai's mHC and @yifanzhang_'s DDL independently arrive at the same principles. When major labs converge on your framework, you know you've hit something universal. We will drop the full paper before the weekend is up 👇
1
1
5
2,415
