Saurya Gur

Humanity enthusiast.

[Saurya Gur @sauryagur]

@gitghxst speed has never killed anyone

suddenly becoming stationary, that’s what gets you

[Tim Urban @waitbutwhy]

[read about what’s happening with AI]

ME: Wow we’re actually legit fucked this is so bad.

[go outside]

ME: Wait those people are all acting like nothing unusual is happening. Maybe I was just in a paranoid bubble? Yeah it must be that. Phew.

[read more about what’s happening with AI]

ME: Wait no we ARE fucked. Shit. It’s all those people outside who are in a delusional bubble.

[go outside]

ME: Wait...

[Saurya Gur @sauryagur]

can we please postpone this until after I get to play GTA 6?

[Saurya Gur @sauryagur]

this is me but still unemployed

[Avinash @avinxshKD]

🇺🇸: I'm unemployed

🇮🇳: 4x hackathon winner | 2x intern | Open Source Contributor

[Saurya Gur @sauryagur]

@ImLunaHey toml is the only right answer

[Saurya Gur @sauryagur]

@gitghxst Given the NSA backdoors in most Intel hardware I don’t think that’s gonna be enough

[Rutger Bregman @rutgerbregman]

Unfortunately, the people who sound crazy about AI have a far better track record than the people who sound reasonable.

New post -->

[Saurya Gur @sauryagur]

@hwchase17 swarm engineering

[Saurya Gur @sauryagur]

I, too, like chinese models

[timo @timo_rf]

JENSEN! THATS NOT WHAT WE MEANT BY CHINESE MODELS!!!!!!!

[Saurya Gur @sauryagur]

@ai_sentience ban banning banning recursive self improvement

[Saurya Gur @sauryagur]

@ashxhart No way this is real

[ice @penguin_pmkt]

the world as we know it could very well end in the next 18 months and there are probably <2m people on earth who comprehend the extent of it

[Saurya Gur @sauryagur]

@_somaxsoma What do you mean by “covert underperformance”?

[Saurya Gur @sauryagur]

We need Indian open models!

[Shalev Lifshitz @Shalev_lif]

So, this isn’t good.

This is very very bad.

If we don’t figure out ways to democratize access to U.S. cyber models that are competitive with the open Chinese models, we’ll end up relying on foreign models whose training pipelines we can’t trust. Defenders will never catch up and we’ll be at even greater risk of AI insider threat.

We don’t have to open up the frontier of closed U.S. capabilities, but there is no reason why the U.S. can’t lead in open-source.

Let the spice flow.

[Saurya Gur @sauryagur]

y’all got any of that 8B

[Qwen @Alibaba_Qwen]

Qwen4 is coming.

Deeper knowledge. Richer world understanding. Stronger coding and agentic capabilities.

Built for real-world tasks — from complex coding and knowledge work to front-end builds and professional software workflows.

Qwen4-Max is now in preview on Alibaba’s Token Plan. You can also get early access to the new model on QwenWork, Qoder, Qwen APP, and QwenStudio.

Put it to the test on a real problem. Show us what you build. 👀

Token Plan: https://t.co/LOB2Jdt3Sk

[Saurya Gur @sauryagur]

@georgeing

[Saurya Gur @sauryagur]

I WAS WRONG

[Saurya Gur @sauryagur]

does that mean if we put brain chips in people and influence their dreams while they sleep we can permanently fine tune their behavior just like we do with LLMs

[Saurya Gur @sauryagur]

If sleep in humans is a kind of memory consolidation, what if dreaming is the brain’s equivalent of offline generative replay; running forward simulations over reactivated and recombined experiences to reinforce learned representations?

[Google Research @GoogleResearch]

Today on the blog, learn about our new instance of a “sleep” phase in continual learning, which enables LLMs to continuously integrate new facts and enhance reasoning while mitigating catastrophic forgetting or static knowledge cutoffs.

More →https://t.co/5v9w5Pn4dw

[Saurya Gur @sauryagur]

@celestepoasts sft is yellow and rl is maroon

[Saurya Gur @sauryagur]

@Dr_Singularity