This can be correct, if they’re talking about training smaller models.

Imagine this case. You are an automotive manufacturer that uses ML to detect pedestrians, vehicles, etc with cameras. Like what Tesla does, for example. This needs to be done with a small, relatively low power footprint model that can run in a car, not a datacentre. To improve its performance you need to finetune it with labelled data of traffic situations with pedestrians, vehicles, etc. That labeling would be done manually…

… except when we get to a point where the latest Gemini/LLAMA/GPT/Whatever, which is so beefy that could never be run in that low power application… is also beefy enough to accurately classify and label the things that the smaller model needs to get trained.

It’s like an older sibling teaching a small kid how to do sums, not an actual maths teacher but does the job and a lot cheaper or semi-free.

I also think heating everything up is the smoothest solution. But to offer an alternative, I’d use dental floss to get in between the bowl and plate. If the bowl has slightly rounded edges (I believe it will), it won’t be too hard to get floss in. With the floss you’ll get inevitably some air in… Which will equalise the pressure and break the vacuum.

As an inferior alternative to floss, fishing line could work for this approach as well.

“There are some bad things on the internet”

“Just… Don’t use the internet?”

I bought a Ricoh GR III (not a GR III digital, an actual GR III) off Facebook marketplace for £350 a few years ago (at this point, the GR IIIx hadn’t been released). It was cheaper than most because it was missing the front decorative ring. I added that back for £10-15, though.

In the year or two that I had it, I realised it wasn’t right for me. It was a lovely camera, but I hated the focal length, the autofocus system always seemed to pick up some random object I didn’t care about, and manual focus with the buttons was between impractical and downright unfeasible. I put it on eBay with an auction start price of £150 or so… And it fetched £450.

Another year or two later I was thinking of getting a GR IIIx as I’d probably enjoy the focal length more. I looked at eBay and damn, both the IIIx and normal III are selling for £800+, with only some poor-condition outliers below £700.

Apparently ricoh’s manufacturing hasn’t kept up with demand and their used prices are absolutely bonkers. I don’t think this has happened to Sony (although I wish, as I need to sell my A7ii soon). I think its a bit of a case-by-case basis.

The ads for apps, Xbox games, trial versions of Office preinstalled, the minesweeper and solitaire collection that are preinstalled but actually ad supported or non-free, depending on the region spotify/TikTok/Facebook also come preinstalled, “Movies & TV”, Bing/MS News…

I think all of those count as bloat. I haven’t included Edge because I guess having a browser is a necessity, or copilot/cortana because you said “excluding AI features”.

Oh no! They’re using an emulator! I choose you NINTENDO! Use “Sue for copyright”!

Unfortunately, it’s not very effective (Anthropic’s type is “AI Company”).

I’ve only started using Storygraph recently (which I also like) but I’d consider a federated alternative. Does anybody know whether its possible to migrate the history from SG to Bookwyrm?

Excellent in which specific sense? Most competitors offer better everything (performance, range, build quality) for a given price point.

The fact that Tesla has managed to make EVs that consistently rank below most ICE brands in terms of reliability is mind blowing.

Over the past 5 years, I’ve installed ubuntu about 30 times on different computers. Not once has an install on an SSD taken me more than an hour, with it typically taking me 30 minutes or less except for rare occasions where I’ve messed something up.

PREGANANANT

It’s the other way around, an Apple Silicon Mac would be able to run an intel binary through Rosetta (I think there’s almost no exceptions at this point). It’s intel macs that can’t run Arm specific binaries.

I thought a few days ago that my “new” laptop (M2 Pro MBP) is now almost 2 years old. The damn thing still feels new.

I really dislike Apple but the Apple Silicon processors are so worth it to me. The performance-battery life combination is ridiculously good.

Less conveniently while costing something like $700 plus a monthly $25 subscription.

I don’t get how it got pitched either.

Also because, as a person who has studied multiple languages, German is hard and English is Easy with capital E.

No genders for nouns (German has three), no declinations, no conjugations other than “add an s for third person singular”, somewhat permissive grammar…

It has its quirks, and pronunciation is the biggest one, but nowhere near German (or Russian!) declinations, Japanese kanjis, etc.

Out of the wannabe-esperanto languages, English is in my opinion the easiest one, so I’m thankful it’s become the technical Lingua Franca.

It’s UE in Spanish, from Unión Europea. (Non-doubled letters because it’s a single Union, there’s no plural like in “States”).

Sometimes people in Spain do use the English acronyms for both EU/USA, but I don’t think I’ve seen it often. Both UE and EEUU are more common from what I’ve seen, and also people rarely say these out loud, it’s exclusively a written language problem.

Well, it has 10x more speakers than German, but it still has fewer speakers than English and most of them are localised in a single country.

I’m talking about running them in GPU, which favours the GPU even when the comparison is between an AMD Epyc and a mediocre GPU.

If you want to run a large version of deepseek R1 locally, with many quantized models being over 50GB, I think the cheapest Nvidia GPU that fits the bill is an A100 which you might find used for 6K.

For well under that price you can get a whole Mac Studio with those 192 GB the first poster in this thread mentioned.

I’m not saying this is for everyone, it’s certainly not for me, but I don’t think we can dismiss that there is a real niche where Apple has a genuine value proposition.

My old flatmate has a PhD in NLP and used to work in research, and he’d have gotten soooo much use out of >100 GB of RAM accessible to the GPU.

If it’s for AI, loading huge models is something you can do with Macs but not easily in any other way.

I’m not saying many people have a use case at all for them, but if you have a use case where you want to run 60 GB models locally, a whole 192GB Mac Studio is cheaper than the GPU alone you need to run that if you were getting it from Nvidia.

So the lack of apple-branded AI Slop is slowing down the sales for iPhones but not for Macs?

Edit for clarity: I’m aware sequoia “has” apple intelligence but in a borderline featureless state, so it’s as good (or as bad) as not having anything.

I recently discovered dupe.com , which works quite well by giving it an Amazon link and finding it on potentially cheaper, definitely non-Bezos websites.