hurrdurr57

The Phi models always seem to do really well when it comes to benchmarks but then in real world performance they always fall way behind competing models.

Will it generate racially diverse Nazis at higher resolutions than their old model?

>the biggest indication of flawed leadership is a company or agency leadership photo where the majority percentage of the people in it are all the same skin tone.

Does this opinion come from your actual experience or just from your ideological indoctrination?

Virtually all non-western businesses have zero concern about fostering racial diversity, they are all failures in your opinion?

Realistically, I think it would have to be from the release date.

Given how quickly AI is progressing from the software side, and how poorly AI scales from just throwing raw compute time at a model, I don't see a company holding onto the lead for very long with that strategy.

If I can come out with a model a year later, and it can provide 95% of the performance while costing 10% as much to run, I think I would end up stealing a lot of customers before they had a chance to break even.

Take Llama3-8B for example, this is an 8 billion parameter model from 2024 that performs about as well the the original ChatGPT, a 175 billion parameter model from 2022. It only took 2 years before a model that can run on a desktop could compete with a model that required a data center.

Well, I guess the question I have is, what exactly does he mean by the "cost to train"? As in, just the cost of the electricity used to train that one model? That seems really excessive.

Or is it the total overall cost of buying TPUs / GPUs, developing infrastructure, constructing data centers, putting together quality data sets, doing R&D, paying salaries, etc. as well as training the model itself? I could see that overall investment into AI scaling into the tens of billions over the next few years.

>Cruz cannot give any interviews without his permission

That's a really weird settlement.

Hammers don't kill people, people kill people

... with hammers

Well, the statement that GPT-4 is 1.8T parameters is a little misleading since it's really a 8 x 220B MoE (according to the rumors at least).

Also the size of the model itself isn't the only factor that determines performance, LLama 3 70B outperforms LLama 2 70B even though they have the same size.

This would be doable, but expensive. You'd need to pay people to go through 10,000 pornographic images and manually blur out all of the fun bits.

Personally, I think hierarchies are tyrannical and oppressive and should be abolished entirely.

As I have just demonstrated my selfless pursuit of equality for all, naturally I should be placed at the very top of the new non-hierarchy.

And their only hope is to get more investors to prop them up.

So it kind of makes sense that they would go with a conservative model that was virtually immune to the AI porn / deepfake panic.

This is what I've heard as well. They started with one cell, when it failed internal expectations they switched to two cells, then when they heard the rumors of what Microsoft was doing with the Xbox 360 they ran to Nvidia to get a real GPU.

>A less charitable interpretation is that they are attempting specifically to recruit people who would prefer to work exclusively amongst people very similar to themselves.

This is the case with pretty much everyone.

When it comes to companies that have an aggressive DEI policy, you may end up with employees more diverse in terms of race and sex, but they are usually all required to adhere to the exact same ideology. They only hire people very much like themselves, only on a different metric.

>DEI is fundamentally anti-marxist.

It's not explicitly Marxist, but come on, it does borrow quite a few key ideas.

That's kind of like saying Methodists are fundamentally Anti-Christian because they've deviated from Catholicism.