While time machine used to be snappy on HDD it was build out of one of the weirdest little hacks hard links in folders. The new Time Machine might front a similar front end but under the hood its a different beast. You're having to mount snapshots as you move through time.
Most UX designers seem to come from a graphic design direction, and they generally do a better job than developers - but not as good as the old school people who did research and wrote detailed guides and such.
The research and guides went out of fashion because actually methodically studying users costs money. I worked at a place that allegedly prioritized design and was really trying to launch more redesigns but also somehow kept losing positions for designers.
It’s time and labor intensive and you generally need to compensate focus groups for their time. Sometimes that overlaps with laziness or apathy, but who hasn’t worked on something and been told the business can’t afford the time or money to do it right?
I'd forgotten there used to be a whole discipline called HCI (Human-Computer Interaction) which seems to have largely disappeared now, at least in a commercial context
I am reminded of the HCI lecturer at university whenever I use something other than my usual KDE desktop.
With a maximized or suitably positioned window, pushing the mouse all the way to the right of the screen activates the scroll bar. To the top right activates the close button.
Small things, but they become annoyances when they are missing.
At least in theory (TypeSafe has been pretty close-lipped about the details so this might just be hot air, and I think the evidence is a bit spotty) this is false, since they use a different reinforcement training method.
If the word “calibration” in probability doesn’t mean anything to you, the difference isn’t very apparent, but that doesn’t mean it doesn’t exist.
It's calibrated where it's calibrated and not where its not. Their "confidence" score is rather bunk, so it's hard to say. As with other LLMs, you have to eval it for your setting, and if you find it calibrated, then great.
Yup, one output token, and read the logprobs of the posible tokens that could have been thos 1st token. Plenty of systems already do this; Typesafe's fundraising and marketing just made it visible.
Good to see interest broadening beyond "just extend thinking." More approaches in the toolbox means fewer problems get treated as nails.
Maybe I oversold the fun-ness of it. The most common failure mode is that it goes into loops and you come back to find it exhausted the output length without getting anywhere.
Such a wonderful take on Firefox. The only thing it is really missing to make it perfect for everyone is proper sync not just on one computer but across systems.
I wish I new of more gems like this. Company sites that go into detail where they went wrong and how they turned things around make for both amazing reading and great educational material.
With all respect, I just downvoted you because you're becoming recognizable. Your account is scarcely a year old with almost no karma, and checking your profile right now, the first page of comments is all but one you posting a link to your own app. You gotta slow down. This is a discussion board, not a growth hacking mixer. Plenty of people become known for popular products and companies they're associated with, but it can't be all you ever talk about. The world at large isn't a sales conference and doesn't find that interesting.