mist

You and I have a rendezvous with destiny.

[🍓🍓🍓 @iruletheworldmo]

‼️huge ssi news.

ilya is about to take his first tentative steps out of the age of research and back into the age of scale. it’s time to smell what ssi is cooking.

ssi have built a small reasoning engine that can compete with much larger training runs because his data is better curated to meta learning. but, more importantly.

we��re about to step into the era of TTT (test time training. gradient descent happening in real time to solve your problems). so instead of a context window you get actual learning.

and because it’s so sample efficient it can be trained on hard to verify tasks that other paradigms can’t touch. everyone else’s weights are frozen, they struggle out of distribution. ssi have created something that has a bundle of knowledge but can truly learn in real time and use that to your advantage.

current approaches are trying to hack their way to ‘learn’ with memory tricks, this thing will updates its weights, remember key lessons, and finally feel like a human level reasoner. this is a huge paradigm shift from the king. early results are very impressive. we can stop watching memento on repeat.

it’s learning all the way down, the descent is real.

[mist @GottMitUns01]

@full_kelly_ @scaling01 data >> compute >= talent

[SpaceXAI @SpaceXAI]

Introducing Grok 4.6.

It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price.

[mist @GottMitUns01]

@YYuke2333 北京有些贵族学校的确��这样，也不会存在严查这类

[mist @GottMitUns01]

@YYuke2333 @vildanden_ai qwen为了刷分overthinking极其严重，基本属于难用的一档。
除了一帮💩都能说成香的立场先行复读机open source ai bro和实在没办法必须要，必须要靠小模型水文章的researcher外没人会用

[mist @GottMitUns01]

@YYuke2333 ��的观点是打不过oa两家的都要通通开源。不是frontier闭源无然后意义
开源这个生态位会一直存在。

[mist @GottMitUns01]

holy shit

[Anthropic @AnthropicAI]

We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis.

It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis from 41.6% to 67.2%.

https://t.co/aZDvqqhHRi

[Anthropic @AnthropicAI]

We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis.

It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis from 41.6% to 67.2%.

https://t.co/aZDvqqhHRi

[mist @GottMitUns01]

@wafuunai @akatsuki__kaze 我有朋友在国内做���器人算法也有好几个朋友在美国读robotics phd。也看过一些vla和wam的paper。
个人观点是robotics目前是个巨大的泡沫，现在这种火热不是硬件支撑的，而是大模型发展衍生出来的通用智能叙事。两边的学术个人觉得都差不多，这个领域���形成一套真正落地的范式还差的太远

[Lisan al Gaib @scaling01]

maybe the nueralese thing is too strongly formulated

but I think the incentives in multi-agent RL are stronger than in single agent RL

and as we scale time horizons and become more safety concerned this seems like something that could emerge

it's still not super likely imo without the right (wrong) environments and pressures

[mist @GottMitUns01]

@YYuke2333 普大帝已经没有了停止战争的能力，寡头私兵组建起来岂是能容易解散的？不去乌克兰前线在国内马上变身无数争夺利益的小军阀。学蒋委员长把非嫡系放到前线送死，一边向维尼索要援助是唯一可行的策略

[mist @GottMitUns01]

@YYuke2333 俄罗斯基本可以说已经入土了，俄罗斯真能爆产能普大帝也不必急着在22年开战。真能爆产能或者自认为能爆产能持久战能赢不可能做一次性梭哈的决策

[mist @GottMitUns01]

@wafuunai @akatsuki__kaze 无论国内怎么说清北c9的最好学生仍然跑北美。
26fall我们这个垃圾学校有个robotics的组面了北大第一名，把大家都惊呆了

[mist @GottMitUns01]

@larsson_za84235 @Goldwater8964 研究的还是胡赛尔，非常值得敬佩，比一团浆糊的下游性别研究强不知道多少

[mist @GottMitUns01]

@scaling01 Another example is bytedance, they have the best resource and the size of talents in china.
They are pretty good at training image/video gen model. Why they are lagging so far behind other labs? I think there might be some legal issue that drive them to avoid distillation

[mist @GottMitUns01]

@scaling01 Nothing to do with distillation. So the real gap can be found by compare deepseek v4 pro (they might release an official version soon) and the OpenAI/anthropic ones.

[mist @GottMitUns01]

@scaling01 Deepseek is the Chinese lab with the least amount of distillation.
Although one of my friends told me a renowned professor in llm extraction attack had a strong evidence that shows ds r1 distilled o1, I do believe they are the best Chinese lab and their feats almost have

[mist @GottMitUns01]

@scaling01 I believe google might have at least a k3 level model. But it is not good enough for them to release and name it as Gemini 3.5 pro. For lab like deepmind, people have a high expectation. If Gemini 3.5 pro could not pair with fable 5, it would break their narrative and reputation.

[mist @GottMitUns01]

@SomethingOnSnow @astridwilde1 they might find some way to access mythos

[levent @__alpoge__]

hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final

((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)