They couldn't even confidently say, "improved", or "made easier".
Don't get me wrong: I'm glad they are at least responsible enough to dodge those lies. I just wish they would put less effort into being "the news", and more effort into accurately describing the features of their model to the general public.
A good set of evolving screenshots is nice progress in development.
It is, but that's not what they were referring to.
With just a little other investigational programming on the side, everything in their presentations would have been truthful, to the level of detail that a customer cares about.
Which is a big reason why I think it's unforgivable that they chose to lie instead. They crossed a brightline in their relationship with their customer without any real benefit from doing so.
- quantitative information about the new system's power relative to its predecessor ("13 times more powerful"; I assume they mean something like "13x faster on typical workloads")
- quantitative information about its speed in absolute terms ("more than 16,000 trillion calculations per second")
- quantitative information about what it will let them do that they couldn't before ("forecast updates every hour", "very high resolution (300m) models", and few particular projects they intend to use it for)
- a little more tech-spec information, albeit stupidly presented ("120,000 times more memory than a top-end smartphone", which I presume means either 120TB or 240TB)
- how much they're spending
- how long they expect it to take to get it set up and working
and they do give the weight in tonnes as well as in double-decker buses. I don't really see that there's much of a problem here.
Regardless of how eloquently they describe the .envs "problem" I was completely thrown off by their "we're doing it better" conclusion at the end. And with zero explanations on how their service is any better.
I left the article questioning myself the validity of their previous claims, knowing that all along I was reading a self promotion piece. Left a sour taste in my mouth. A disclaimer at the top would have been the proper thing to do...
“I don't think they were ever upfront about what their device can really achieve before.”
They’ve been very upfront about this. The problem is all the noise that’s been generated by armchair critics who have spammed every article about them calling them a scam/vaporware/the next Theranos, etc.
I’d genuinely expect the people who built and operate the thing to have a far better write up than what amounted to “it no worked lol”. Sure, NN’s are opaque, but this org paints itself as the herald and shepard of ai and they just produced a write up that’s hardly worthy of a primary-school-child’s account of their recent holiday.
I missed that blog post somehow, thanks for sharing that. Bit disappointing that it's just the 7B model, it's a good starting point for fine tuning another small model, but it really isn't useful on its own.
> This is just a press release, and not a great source of in-depth info
Not sure I'd call outright lying/getting the most basic points wrong "not a great source of in-depth info", like the "local-first" part.
> Perhaps offer more clarity in future to avoid such confusion.
Sorry, I'll be more explicit: I give absolutely no credence to what a manufacturer "expects" or "hopes" or "strongly implies" their product to be able to do in a thinly-veiled press release, especially when they're unwilling to give any concrete facts to back those hopes up.
It's part of the marketing. By saying their models are powerful enough to be gasp Dangerous, they are trying to get people to believe they're insanely capable.
In reality, every model so far has been either a toy, a way of injecting tons of bugs into your code (or circumventing GPL by writing bugs), or a way of justifying laying off the writing staff you already wanted to shit can.
They have a ton of potential and we'll get there soon, but this isn't it.
The problem is that they've been making lots of promises but then also publishing misleading demos that seemed to disobey the basic laws of physics (i.e. demos showing subtractive image synthesis in a compact form factor) which makes it hard to take any of their tech claims seriously.
It purported to be more accurate, not 100% accurate.
I wanted to reuse the language in the first post to make the broader point that there is a difference between a company which has already been doing something for 15 years and a company which has never done that thing but could feasibly do it in the future.
As usual, they give us a press release and two or three examples, but no real data or systematic analysis about its consistency, which is the real problem with any of these models.
I'm excited about the potential of things like this, I just don't like this wish-wash way of presenting a technical tool.
What makes you sure of this? Does something in their execution thus far indicate "little to no understanding of what they need or how to improve their product"? To me it seems quite the opposite, so I'm curious what you are seeing.
They couldn't even confidently say, "improved", or "made easier".
Don't get me wrong: I'm glad they are at least responsible enough to dodge those lies. I just wish they would put less effort into being "the news", and more effort into accurately describing the features of their model to the general public.
reply