Live data from Hacker News

Emad Mostaque resigned as CEO of Stability AI

stability.ai

341–350 of 429 posts

Re: Emad Mostaque resigned as CEO of Stability AI

#341

Earlier quoted context omitted.

Lack of data parallelism is implied by computation that is performed. You gradient descend on your state. Each step needs to work on up to date state otherwise you're computing gradient descend from state that doesn't exist anymore and your computed gradient descent delta is nonsensical if applied to the most recent state (it was calculated on old one, direction that your computation calculated is now wrong). You als…

Your gradient descent is an operation on a directed acyclic graph. The graph itself is stateless. You can do parts of the graph without needing to have access to the entire graph, particularly for transformers. In fact this is already done today for training and inference of large models. The transfer bottleneck is for currently used model sizes and architectures. There's nothing to stop you from building a model so…

Your DAG is big. It's stateless for single pass. Next one doesn't operate on it anymore, it operates on new, updated one from previous step. It has fully connected sub DAGs.

There is nothing stopping you from distributing assembly/machine code for CPU instructions, yet nobody does it because it doesn't make sense from performance perspective.

Or amazon driving truck from one depo to other to unload one package at a time to "distribute" unloading because "distributing = faster".

Re: Emad Mostaque resigned as CEO of Stability AI

#342
post #292

Looks like he went to crypto https://x.com/sreeramkannan/status/1771340250801127664?s=46

EigenLayer's tag line is? "Build open innovation ∞ play infinite sum games. Also: @eigen_da" (from their twitter profile). Infinite sum games...

LMAO

"infinite games" + "positive sum" => "infinite sum"

has big Sarah Palin energy: "refute" + "repudiate" => "refudiate"

https://www.npr.org/sections/itsallpolitics/2010/11/15/13133...

Re: Emad Mostaque resigned as CEO of Stability AI

#343
post #277

Earlier quoted context omitted.

Lack of data parallelism is implied by computation that is performed. You gradient descend on your state. Each step needs to work on up to date state otherwise you're computing gradient descend from state that doesn't exist anymore and your computed gradient descent delta is nonsensical if applied to the most recent state (it was calculated on old one, direction that your computation calculated is now wrong). You als…

Openai had a more distributable algorithm: https://openai.com/research/evolution-strategies But given that they don't seem to have worked on it since, I guess it wasn't too successful. But maybe there is a way

Yes, if there was something interesting there you'd think since 2017 something would happen. Reinforcement Learning (that is compared with) is not particularly famous for its performance (it is it's biggest issue and reason for not being used that much). Also transformers don't use it at all.

Re: Emad Mostaque resigned as CEO of Stability AI

#344

6 months ago, I expressed my doubt about the viability of Stability business model on HN News. Mostaque answered the questions with a resounding yes and claimed the business of Stability.AI is better than ever. Today he resigned.

He still owns a ton of stock, and last month said they were on track to be cash flow positive this year.

Re: Emad Mostaque resigned as CEO of Stability AI

#345

Earlier quoted context omitted.

People love illustrator

It was good in the 00s but now it’s rickety and antiquated and the GPU acceleration was never implemented correctly. Figma could build an illustrator killer in 6 months if they wanted to and it would be obliterated. If they actually tackled this task people would be kicking themselves for putting up with the shambles that is illustrator for this long.

> Figma could build an illustrator killer in 6 months if they wanted to and it would be obliterated

Statements like this are almost always wrong, if for no other reason that a technically superior alternative is rarely compelling enough by itself. It that weren’t the case you would see it happen far more often…

Re: Emad Mostaque resigned as CEO of Stability AI

#346
post #18

The underlying reason is that current AI businesses and models are failing to capture any significant economic value. We're going to get there, but it will take some more work. It won't be decades, but a few more years would be helpful

90% of startups fail, so it's not just a matter of waiting for the tech to get better and having more value - most of the current players will simply fail and go out of business.

Re: Emad Mostaque resigned as CEO of Stability AI

#347
post #140

Earlier quoted context omitted.

The `big' AI models are trillion parameter models. The medium sized models like GPT3 and Grok are 185b and 314b respectively. There is no way for _anyone_ to run these on a sub $50k machine in 2024, and even if you can the token generation speed on CPU is under 0.1 tokens per second.

ChatGPT is 20B according to Microsoft researchers, also the fact that big AI models are trillion parameter models is mostly speculation, about GPT-4 it was spread by geohot.

GPT-3 was 175B, so it'd be a bit odd if GPT-4 wasn't at least 5x larger (1T), especially since it's apparently a mixture of experts.

Re: Emad Mostaque resigned as CEO of Stability AI

#348

Earlier quoted context omitted.

> Many/most of the GenAI companies have interesting ideas but no real business plan - This is most likely the reason being the SAMA firing, to be able to re-align to the MIC without terrible consequence, from a PR perspective. no criticism here aside from the fact that we will see the AI killer fully autonomous robots will be here and unfettered by 'alignments' much sooner than we expected... And the MIC is where all…

> no criticism here aside from the fact that we will see the AI killer fully autonomous robots will be here and unfettered by 'alignments' much sooner than we expected... What exactly do you mean with this sentence? That less woke/regulated companies will suddenly leapfrog the giants now? What timeframe are we talking here? And do you mean for example non public mil applications from US/China or whatever or private m…

maybe they are right now?

Aren’t they? Pretty sure that most tactical and strategical decisions are automated to the bottom. Drones and cameras with all sorts of CV and ML features. You don’t need scary walking talking androids with red eyes to control battlefields and streets. The idea of “Terminator” is similar to a mailman on an antigrav bicycle.

Re: Emad Mostaque resigned as CEO of Stability AI

#350
post #32

He's leaving to work on decentralized AI? That's exactly what Stability AI was doing before it became clear the economics no longer work out in practice, and starting a new company wouldn't change that. (Emad is an advisory board member to a decentralized GPU company, though: https://home.otoy.com/stabilityai/ ) Obviously this is the polite way to send him off given the latest news about his leadership, but this rati…

Think seti at home. Instead of wasting all the compute on bitcoin we pretrain fully open models which can run on people's hardware. A 120b ternary model is the most interesting thing in the world. No one can train one now because you need a billion dollar super computer.

For the record, this already exists in the open source world for Stable Diffusion.

https://stablehorde.net/

You can host a local horde and do "Seti at home" for stable diffusion.

Post reply on HN