1,048 karma · joined February 10, 2022
It seems like "some people love it" would be the MUCH more interesting (if not disturbing) story.
Learning isn't fun doesn't really relate to the fact that bodily functions exist. If a kid is well but going to the restroom for 30 minutes every class period and on their phone etc just skipping.... I guess punish that as if it were skipping? You can't just assume negative intent across the entire student body when it comes to checks notes pooping in a toilet like a person?
If the brackets are 10,000 @5%; 100,000 @10%, if you make 100,000 your first 10,000 is taxed at 5% and the next 90,000 is taxed at 10%.
Outside of long/short term gains, all things being equal, the timing is not especially important
We’ve got a fishy death, no documentation on the door plug, and more incidents.
All the extra stuff is great, but the original is the perfect album.
What music do you typically like? There will definitely be a gateway to jazz in there somewhere.
My project pre-dated the composition API and some other bells and whistles that I've never looked into since leaving FE projects... but IMO, a really good framework with a good community and lots of resources to learn.
It’s crazy to expect all the music in the world on demand for $12 a month or whatever. Spotify can probably do better in some ways but I don’t see how any of this justifies not giving the artists the pennies your mentioning.
People concerned with data quality from LLMs should really see the inconsistencies we came up with!
I couldn't be farther from a math nerd.... I avoided it as much as I could throughout school.... but things like that are just so interesting and weird. How on earth (and for what reason) did they find this out? The properties of this number are interesting enough but the process to discover it is just so crazy.
Otherwise, the (mostly) static nature of Tesla models kept the used Tesla market in decent shape
now if he sold a boatload of stock before a horrific earnings report... thats another story
If you feed something like a bad transcript or some other very lossy (but not strictly scrambled) input.... it really can roll with it and just spit out correct information. Bad tokens in don't necessarily mean bad tokens out....
I'm sure there is a limit to how many tokens can be flat out bad before the "next token" in the response is thrown off, but after seeing what it can do with some of these inputs, the fact it can unscramble is not at all surprising/interesting.
Facial features and fingers just.... are hanging off the body in extremely non-uniform ways w/ no real set proportions. It isn't totally intuitive to me why its so bad at it, but faces especially are so unique and the musculature of the face is so fine.... learning a representation must just be really really difficult.