Dax

It's not really practical. I mean, it's like a reduced version of the model, and it's very slow, and the hardware requirements are pretty crazy. So no, you can't.

It's not really practical. I mean, it's like a reduced version of the model, and it's very slow, and the hardware requirements are pretty crazy. So no, you can't.

Is it commercialized in any way? There's like a hosted one from the company, but it's in China, so people feel sketched out by that. But then it's been re-hosted because open source has been re-hosted by a bunch of providers that you're familiar with. Like Cloudflare, I think, has a version of it.

Is it commercialized in any way? There's like a hosted one from the company, but it's in China, so people feel sketched out by that. But then it's been re-hosted because open source has been re-hosted by a bunch of providers that you're familiar with. Like Cloudflare, I think, has a version of it.

There's been a few others. I don't think it's good. Oh, really? Well, it's like not better than anything else. It's just a... recreation of what's already there.

There's been a few others. I don't think it's good. Oh, really? Well, it's like not better than anything else. It's just a... recreation of what's already there.

I don't believe any... I mean, it's just like... None of the information about it is true.

I don't believe any... I mean, it's just like... None of the information about it is true.

I just heard it's cheap. They're claiming they did... They're claiming they trained the model for $5.5 million, which is like... a crazy uh man like several orders of magnitude less than what OpenAI's models costs. Everyone was like dunking in OpenAI.

I just heard it's cheap. They're claiming they did... They're claiming they trained the model for $5.5 million, which is like... a crazy uh man like several orders of magnitude less than what OpenAI's models costs. Everyone was like dunking in OpenAI.

So the reason, it's very noisy. There is true interesting things that they did. Like, so you can't take that away from them. Like, it's impressive. But that doesn't mean what they're saying about how it was done is true either. The numbers are just like way too much of a lie. Like, there's no way that one, they're that low. Two, there's a lot of reasons for them to make it up. Right.

So the reason, it's very noisy. There is true interesting things that they did. Like, so you can't take that away from them. Like, it's impressive. But that doesn't mean what they're saying about how it was done is true either. The numbers are just like way too much of a lie. Like, there's no way that one, they're that low. Two, there's a lot of reasons for them to make it up. Right.

Well, China's not allowed to have certain DPs. What? Because of the export.

Well, China's not allowed to have certain DPs. What? Because of the export.

Because of the export controls.

Because of the export controls.

On paper, NVIDIA is not allowed to export... certain levels of GPUs to China.

On paper, NVIDIA is not allowed to export... certain levels of GPUs to China.

Yes.

Yes.