This seems to be a common denominator for what LLMs actually do well: Finding bugs and explaining code. Anything about producing code is still a success to be seen.
I personally know that L’oreal will buy back and destroy products of theirs from outlets, just to keep the prices up. These items are often bought in bulk on grey markets by discount outlets. Not only does L’oreal destroy the products, they pay for them to do so. None of this is shocking IMO.
We have (in Denmark) a lot of shows on our public broadcasting network DR, and similar for NRK (the Norwegian counterpart) that are meant for toddlers, many of which share these traits:
- Slow, meaning long scenes without cuts.
- Calm speaker narrating something happening in simple words and with repetition.
- Babies like to watch other babies, and toddlers other toddlers.
Go for older shows (90s, 2000s). Once they watch newer stuff which has a faster pace, it’s hard to get them to watch the oldies.
Once they get older (5+), I would recommend Stillwater on Apple TV.
Unless you have a zero screen policy with your kids - which I can definitely recommend - the trick is to give them incredibly small doses and being consistent with it. My son is 6 and he is perfectly happy with 30 minutes once a week when his younger brother is taking his nap.
We tell him it’s not good for him, which is why we limit it. He occasionally complaints, but most of the time he is looking forward to the 30 mins, and will close the iPad on his own when the timer ends. Consistency is key.
Second advice is to not give them authority over what to watch. No YouTube. Spend some time curating what will be acceptable. In the first few years (we started allowing it when he was 3) they will watch anything. Just stay of the dopamine stuff!
It helps that we never have the tv on when they are awake. Break your own bad habits first - but that’s general parenting advice…
If I see this many questions that could have been taken straight from an FAQ, all followed up with “great question, we do so and so..”, my alarm bells go off. Shady marketing indeed.
We use their earlier Mixtral model because it outperforms llama for our use case. They do not release full models for marketing purposes, though it definitely grabs attention!
You may need to revise your views..
I think it’s a fair comment. A lot of readers on HN are adept debuggers, and will start to analyze everything from the first paragraph. By burying the lede like that, it feels like wasted time, to have begun debugging before the (incredibly important) part about the unusual setup was revealed.
Seems almost implausible that the protagonist, with his technical knowhow, did not think of this earlier..
Anyway, it’s a matter of storytelling, and that matters!
The “browse with bing” feature allows it to fetch a single webpage into the context, but the new cutoff allows _everything crawled_ to be context (up to the new date, that is)
I would assume that the majority of the animals that humans kill are kept in captivity and therefore not exposed to other animals. Since this breeding/captivity is happening at a staggering scale, my second assumption is that animals have absolutely no way of keeping up with humans.
I would also like to see the numbers of kills by non-humans too!
I’m quite appalled actually, seeing it laid out like this, and I eat my share of chicken and fish..
> The author touts Spotify's recommendation engine and rightly so, nothing comes close.
Actually the author does the opposite, saying that YouTube’s algo is superior.
For me, Spotify used to be THE recommendation engine, but the last few years I found it to regurgitate a lot of the same songs, and recommended the same songs to people around me.
To be fair, I’m far less in “discovery mode” than I used to be, but Spotify is contributing to this by no longer inspiring me. It’s easy to get Spotify like everyone else and just be over with it, but a healthy discussion about active choice is always welcome.