201–210 of 217 posts
Don't have iOS. But I do wish that Android with detect the current language you are writing in. I have 3 languages on my keyboard and sometimes it suggests an autocomplete in the wrong language.
I have German, English and Polish and it works perfectly.
Weird
Earlier quoted context omitted.
I can never make iOS write “Hong Kong” when swipe-typing; tried setting up forced text replacement entry (kong->Kong) to zero effect. It is always “Hong kong”.
Weird - just tried it and ir capitalised it correctly. I'm on the ios14 public beta, though.
Earlier quoted context omitted.
When it comes to machine learning, the subtlety here is that there are at least two sides or facets to machine learning: (1) training and (2) inference. It's fair to say that there are multiple areas for AI leadership. It is generally believed that: Model Creation (1) Those with the access to the best (which is not necessarily the most, but often believed to be) data have a strong starting point for training models;…
But can you lead in (2) without being good at (1)? If you train on limited data, then your inferences will be of poor quality, even if they have low latency. So in sum, it seems you can't be a leader in ML without both.
Being a leader in (1) does not mean you'll be good at (2), and vice versa.
There's also a difference between limited data and good enough data.
If you train on good enough data, you can have good enough models.
If people believe the focus of AI/ML should just be precision/recall, or other measures of accuracy, and having tons of data, they're missing many other areas and elements for what make AI/ML successful in application.
Earlier quoted context omitted.
> ... in the era of weapons-grade language models? Just an aside: Although common perception is that "weapons-grade" refers to the highest quality, in reality it refers to the lowest-possible quality that meets the bare minimum requirements!
So? That’s what every standard means: “anything just barely over the minimum that the standard specifies”. That can definitely be wonderful and impressive if the standard is high to begin with. You might as well refer to the bronze medal as “just barely enough to be in the awards ceremony”. Is that supposed to be clever somehow?
I thought the parent comment was mildly interesting. There's no need to be so combative.
Earlier quoted context omitted.
It's a shame that Apple doesn't have an industrial research lab; it certainly has the funding to create a lab that would rival Microsoft Research or perhaps even legendary labs like Xerox PARC and Bell Labs. Apple used to have a lab: during the "interregnum" years of 1985 to 1997, Apple had a fantastic research group called the Advanced Technology Group, led by the late Larry Tesler. In some ways this group was a sor…
Has the culture of secrecy benefited them at all since the launch of the iPhone?
Earlier quoted context omitted.
I have German, English and Polish and it works perfectly.
Not so great with Dutch, French and English though. Weird
It's a hard language and not as good covered usually with ML as English or German is.
Setting aside all the other work Apple is doing in this area, can someone explain why iMessage autocorrect is unacceptably bad (to put it diplomatically) in the era of weapons-grade language models? Its behavior is totally bizarre. It’s like an underpowered chess engine that makes flagrant blunders: capitalizing random words in the middle of a sentence (rock -> Rock), the context sensitivity of an actual rock, forcin…
The swipe keyboard is atrocious too, I simply cannot swipe the word ‘you’ with any reliability. It becomes tut, or töre. Töre. Instead of the second most common pronoun in the entire English language.
Earlier quoted context omitted.
So? That’s what every standard means: “anything just barely over the minimum that the standard specifies”. That can definitely be wonderful and impressive if the standard is high to begin with. You might as well refer to the bronze medal as “just barely enough to be in the awards ceremony”. Is that supposed to be clever somehow?
> Is that supposed to be clever somehow? I thought the parent comment was mildly interesting. There's no need to be so combative.
Earlier quoted context omitted.
> Is that supposed to be clever somehow? I thought the parent comment was mildly interesting. There's no need to be so combative.
Pointing out tautologies that only sound convincing because of misleading phrases, isn't interesting.
FWIW, your comment above (about standards) was also mildly interesting to me. I hadn't really given the phrase much thought.
There was no need to be rude about it though.
Earlier quoted context omitted.
The reason iOS type correction is so piss-poor is, to me at least, because it is not a 'typing' correction mechanism, it is pronunciation correction system. It bases it's auto replace on what the letters appear to be saying or sound like, and NOT, as it should, which would be simply and effectively based on typos on a qwerty keyboard. That is why. And it is like this because Apple want to steal your voice and 'prefer…
The weird part is in my experience their "swift" input works miles better than the regular tapping. It's truly odd. Tapping I have to correct the phone all the time, while using swift (with the same keyboard!) I quite rarely do.