مرد خدا

Hmm, okay....

[مرد خدا @techsavvyl1]

بار الهی اگر برای هر مردمی سپاه‌ی گذاشته‌ای، پس بدان که همین فرزندان بدخشان سپاهیان ما اند و بزرگانشان سپهسالاران ما. خدایا تیرشان را بران کن، عزم شان را مستحکم، قدومشان را ثابت و جنگ شان را کوبنده. تا باشد که خورشید خراسان طلوع کند و ملت‌مان روشنایی بقیه ملل عالم باشد.

[سپاهی میهن @HSFront]

امشب، شبی است که خشم فرزندان آزادی‌خواه میهن بر سر ظالمان فرود می‌آید.
سپاهیان میهن با اراده‌ای پولادین تقاص سال‌ها ظلم و خفقان را پس می‌گیرند؛ هیچ نقطه‌ای برای طالب امن نخواهد بود و شمارش معکوس برای سقوط نهایی‌شان به سرعت به پایان نزدیک می‌شود!

[Sam Altman @sama]

The back of the Navier-Stokes milestone shirt is the last line of the chain-of-thought from the solution:

[Dr Singularity @Dr_Singularity]

Being an optimist doesn't mean thinking everything's great.

It means believing we can fix things.

[مرد خدا @techsavvyl1]

@Dr_Singularity Thats why i see trump is getting younger every single day

[مرد خدا @techsavvyl1]

@behradjaved_com من نصبش کردم و عالی بود

[Sakana AI @SakanaAILabs]

Introducing Multi-Agent Self-Supervision (MASS): Recursive self-improvement through collective intelligence

Blog: https://t.co/LkQe5k7jOW

As AI systems tackle increasingly open-ended research problems, human experts may struggle to assess results and guide further progress. The model itself may be the best available optimizer and evaluator. The challenge is obtaining useful self-supervision within that homogeneous loop, where a model risks reinforcing its own blind spots rather than correcting them.

We ask whether a model can learn from the collective intelligence of its own team.

We introduce Multi-Agent Self-Supervision (MASS). One shared model solves tasks through a team of virtual subagents, searches for better multi-agent workflows, then uses its own judgments to select the best ones.

Training on the best team’s executions provides self-supervision by distilling the team’s collective experience into the shared model.

We ran two cycles using a 27B open-weights model on synthetic open-ended research tasks. After the second cycle, score per output token reached 1.2-1.6x the base model's level across four research benchmarks.

Three key results:

First, MASS enables role generalization within an RSI loop: training only on task-solving data from the optimizee naturally improves the capabilities of both the optimizer and the evaluator.

Second, multi-agent trajectories are more efficient training data than single-agent trajectories for RSI.

Third, optimizer capacity, rather than evaluator capacity, can be the bottleneck in the speed of improvement.

We believe the community needs to investigate homogeneous RSI carefully to support AI safety. When a model is also its own evaluator, shared blind spots could allow mistakes to be accepted as progress and reinforced through training. For AI to improve beyond human expertise safely, we need to understand these failures and preserve effective ways to detect them and intervene.

Paper: https://t.co/SzXDlAzrUO
Code: https://t.co/xc870ZHRgh

This work was a collaboration with researchers at UC Berkeley.

[مرد خدا @techsavvyl1]

@bryan_johnson Are you the anti Christ?

[مرد خدا @techsavvyl1]

@Hesamation Just 4 years ago it was a Sci Fi thing

[مرد خدا @techsavvyl1]

@dhruvalgolakiya Humanity is done.

[مرد خدا @techsavvyl1]

@nutlope Hallo Hassan, I would really love to try your work. Thank you!

[Sankalp @3sankalpsingh]

He already did that
https://t.co/y671d4GtzL (Microsoft Excel)
https://t.co/mOWdjysKQX (PowerPoint)
https://t.co/dBiQA8lnzp(Microsoft Word)

[Brian Roemmele @BrianRoemmele]

In 2001 I said AI was ultimately the end of software and the OS. It got me laughed at.

Welp

“Creator of free Adobe clones unveils open-source Microsoft Office Excel, PowerPoint, and Word replications - Autodesk tools and other subscription-based services may come next”

Soon SI will just make what you need when you need it, save the code if you want and you can call it Excel or Word. Either way the old world is over.

You won’t believe what I am saying now.

[Money Quotes @MoneyQuotesX]

Humans are designed to create. That's why you get depressed when all you do is consume.

[Haider. @haider1]

Ex-Google CEO Eric Schmidt:

"the belief is among some people in San Francisco that there will be an explosion in learning that will accelerate past what humans can understand"

Imagine a mathematical proof that's too complex for any human to understand, yet we can verify it's true

What does it mean if humans can never understand it?

[مرد خدا @techsavvyl1]

@adcock_brett @minhsmind I hope it is not another moving arm robot.

[EP @eptwts]

keeping up with AI advancements is a luxury only AI researchers, AI creators & the unemployed can afford...

[مرد خدا @techsavvyl1]

@TheSpacerr You asked it in what?

[مرد خدا @techsavvyl1]

@hisGenie_ @adcock_brett @tomfgoodwin I agree. It is too humane.

[مرد خدا @techsavvyl1]

@chazza How do you get that much tokens?

[Irakli 🚀 @TheSpacerr]

I asked Claude: “Show me what human life is for you.”
This is what it showed me...🎧

[مرد خدا @techsavvyl1]

@adcock_brett @gauravvohra Hark is badly like a real human.