JFMN

Ascending | Experimenting | File Ops | Booking Agent @amfionhq | Wellness

[JFMN @jayfmn_]

I built Verbaform. Programmable dictation where your voice is the input AND the editor.

You speak. Then you say something like "verba trim fillers" and it edits what you just said. Deterministic, local, instant.

The complete local product is free. It's live today.

[Bill Ackman @BillAckman]

.@MIT What are you doing about this?

[AutisticClips @AutisticClip]

Tony Hinchcliffe praises Elon Musk

“Call me a d*cksucker whatever, I love it. I just think the guy is f’ing unbelievable and we’re so lucky to live in an era where he’s alive.”

[OpenAI Developers @OpenAIDevs]

We built Pando, a new Git backend for our monorepo, to handle more pushes and fetches without changing how our engineers and agents use Git.

Pando cut push latency by ~80% and now serves nearly a petabyte of Git data a day. We can grow Git capacity alongside agent usage.

[Simon Willison @simonw]

There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing https://t.co/6O7WltGP7K

[Matt Mickiewicz @MattMickiewicz]

Over the past 10 days, I became the digital Nick Shirley.

Using public & private records my AI agent swarm found ~$787 million in government billing that made absolutely no sense.

Thread 🧵

[Chubby♨️ @kimmonismus]

I think Chandan is right. This is a significant release, the importance of which can hardly be overstated. It is a giant step toward recursive self-improvement.

[Alex Krentsel @AlexKrentsel]

i've reproduced the results from the MIT blog post today on weight-space programming. but one additional result i have to add: this is a new, step-change emerging capability that's appeared in just the latest set of models.

[Thorsten Ball @thorstenball]

I swear to you: at least one of the replies here will make you say "what the fuck" out loud

[Rikiya Takehi @rikiyatakehi]

@FeinsteinKen @lateinteraction @a1zhang @jayden_teoh_ I think so too. Not just own weights. Any other weights!

[Ken Feinstein @FeinsteinKen]

@rikiyatakehi @lateinteraction @a1zhang @jayden_teoh_ This is clearly how it ends. The machine just recursively optimizes its own weights.

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[Token Gremlin @TokenGremlin]

GPT-6.2 Astra is coming soon, and Sam Altman has described it as one of the most powerful models they've developed internally, capable of solving math problems that even the greatest minds on the planet struggle with.

OpenAI literally created a benchmark to determine whether this qualifies as AGI, and they're planning to make this model accessible to everyone, with the goal of "benefiting all of humanity."

[Palmer Luckey @PalmerLuckey]

@dickclucas What if I want to use the RAM to do things that are prohibited by the terms of service of whoever is running the datacenter?

[Dillon Mulroy @dillon_mulroy]

have your bot, dot, muse, or duet agent monitor this list - adding every single ai slop poster i see to it so you can easily block them

https://t.co/p3aL2EvoA2

[Elon Musk @elonmusk]

@shaunmmaguire @Indian_Bronson I think we can do better

[Andrew Curran @AndrewCurran_]

If you don't have a personal agent filtering out the slop, you will drown.
If you don't have a personal agent running your cyberdefense, you will be hacked.
This is the new divide. Choose open or closed, choose the company you like best, but choose soon.

[JFMN @jayfmn_]

@chetaslua @karpathy GPT 10

[TheManWhoFellToEarth @ReturnofColin2]

This is brilliant.

[JFMN @jayfmn_]

@karpathy Ben Eater is a legend.

[Pedro R. @SeniorManager]

Le preguntaron a Bukele en Argentina cómo sacaron a los jóvenes del modelo del narco: la casa grande, el coche caro el smartphone… cuando trabajar honesto tarda años. Su respuesta duele porque es verdad

"El gobierno llega tarde si espera reinsertar al que ya violó, ya mató y ya cortó cabezas. A un niño de 6 u 8 años sí se le puede cambiar la vida. Al que ya eligió el mal... el joven que vende tomates en el mercado, suda, carga, gana $5… y en la esquina el pandillero se los quita. ¿Qué incentivo gana? ¿Vender tomates o ser el que cobra? Por eso nunca se gana la guerra de incentivos si el criminal sigue en la calle.

Porque el Estado no puede pagarle triple salario mínimo para que deje el crimen. Eso no es gobierno, sino ingenuidad cara. Un organismo me dijo: «no puedes acabar el crimen de un golpe, son 70 mil empleos, esa plata llega a los hogares y se gasta legal». Lo que ellos insinuaban era que dejara vivir al cáncer para no lastimar la economía. Bukele les respondió con otra analogía: si tengo cáncer con metástasis y me ofrecen cirugía, quimio y quedar débil… pero vivo, me opero. No me quedo así nomás esperando que el tumor sea educado."

Le dijeron que el PIB podía caer hasta 10% por sacar esa economía sucia. No cayó. Subió. Porque el miedo también es un impuesto. Y ese sí lo pagábamos todos los días. Sacaron al pandillero de la ecuación y no sólo para castigarlo, sino para que el muchacho que todavía no entra en ese mundo vea otro menú. Ya no es: «me hago pandillero y gano». Ahora es: «me hago pandillero y pierdo mi libertad». Eso no es odio, es orden. Es proteger a la familia que sí trabaja, a la que manda la remesa, a la que nunca tuvo casa grande porque no cobraba renta de sangre. 30 años de discursos. 7 años de resultados. El resto es teoría para intelectuales.

Tal vez necesitamos un Bukele en España. Y en el mundo.