IanCal

58 ms·

901.
by IanCal 2y ago
> Speaking as someone who has experienced homelessness, for a short while, I believe the only practical solution is to give people housing at the outset before the desire/need to move elsewhere takes root. Also national support not
902.
by IanCal 2y ago
The fundamental thing behind SPARC is (iiuc) to do this far faster by being drastically smaller. Iter was designed a long time ago based on materials available at the time. SPARC can be much smaller with much stronger magnets. The argument
903.
by IanCal 2y ago
> And as I said in my original comment they are probably not even able to extract the board state very well, They do measure and report on this, both in summary in the blog post and in more detail in the paper. > otherwise they would
904.
by IanCal 2y ago
Promoting to anything other than a queen is rare, and I expect the next most common is to a knight. Promoting to a bishop, while possible, is going to be extremely rare.
905.
by IanCal 2y ago
> hey don't appear to do the obvious thing which is display the reconstructed "internal" board state. I've very confused by this, because they do. Then they manipulate the internal board state and see what move it mak
906.
by IanCal 2y ago
I also wonder what kind of invalid moves they are. There's "you can't move your knight to j9 that's off the board", "there's already a piece there" and "actually that would leave you in check&quo
907.
by IanCal 2y ago
Huh? This is perfectly fine in the EU. The reason they have the banners is because they need consent. Removing the banner is totally fine.
908.
by IanCal 2y ago
I don't get what the irony is here.
909.
by IanCal 2y ago
You are arguing for a much more complicated set of rules that now have to define what's obvious and what isn't. Is your preferred rule that they don't have to list allergens if they're already in the ingredients? Is it a
910.
by IanCal 2y ago
Outside of the fact that afaik they didn't, I would think this is intended to anyone who may contact the author as it is a personal blog. This is the comment section on a post shared by someone else as I understand it.
911.
by IanCal 2y ago
Not super relevant but your site is super janky for me (chrome, ubuntu). I get a scrollbar for the second part of your site, which then captures scrolling so if my cursor is in the middle of the page and I go down then up I am stuck. https
912.
by IanCal 2y ago
Increased relevant to other providers of different llms. So I'd pick watermarked X over non-watermarked y if x performs better than Y.
913.
by IanCal 2y ago
Anecdata from the other side, I've done a few passports for family now and it gives immediate ratings and we've not had any issue at all.
914.
by IanCal 2y ago
I'm totally happy having huge amounts of my use of llms identifiable as from an llm. I don't see many important cases for me where I need to pretend it wasn't from an llm. I will happily lose those cases for increased perform
915.
by IanCal 2y ago
That's fantastic thanks - the sliding sync part addresses an issue a client I've got would be facing at some point. I'll have to check out the details of that.
916.
by IanCal 2y ago
Does synapse support this at the moment? I can't quite tell from the writing here if/how you can run things that support this - my guess was no because it mentions improvements to the js SDK are required and there's no warnin
917.
by IanCal 2y ago
AI integration is really handy for coding. I wouldn't consider an editor that didn't have decent support for models and getting diffs back.
918.
by IanCal 2y ago
If I make you go through the DMCA counternotice process for buying apples from me, that's not a problem with DMCA, it's me having an absurd process around purchasing apples. I don't see how people weirdly requiring this proce
919.
by IanCal 2y ago
Why? The very nature of next token prediction means it's entirely capable of having that. It's not multiple passes, it's just one pass. You making multiple calls is just inserting fixed tokens then asking it to carry on compl
920.
by IanCal 2y ago
We literally ban people from naming their children with unpronounceable names.
921.
by IanCal 2y ago
That doesn't test things in a useful way, and relies on having an official dataset lie. Good ingestion code should ignore those, and then you're not even testing the frontend of those systems.
922.
by IanCal 2y ago
And assuming all your consumers are not sloppy is impractical. We sanitise input all the time. This is not particularly unique. There isn't a great loss in this restriction of company names.
923.
by IanCal 2y ago
It's nowhere near instantaneous for physical items. I live near a library and it's a drive or a half hour walk away, open a few days a week during shortened times. That's nothing like clicking a button and renting a film for
924.
by IanCal 2y ago
The amount of time makes a practical difference. Books you take out for a week or more usually, games rentals and films (from the olden days when I'd get to go to an actual store) were at least several days. Renting a film is something
925.
by IanCal 2y ago
Per hour of computer execution is a poor measure. Imagine it did this twice as fast, and cost the same. Is that worse ? A per hour figure would suggest so. What if it was far slower, would that be better?
926.
by IanCal 2y ago
Huge seconding of cursor.
927.
by IanCal 2y ago
> as active is taken to mean electrical control of current flow. Does the building of logic gates controlling a motor not show electrical control of current flow?
928.
by IanCal 2y ago
Hang on, can you explain why this is passive and not active? > Harnessing the described phenomenon, we created the first semiconductor-free active electronic devices fully 3D printed via material extrusion. We demonstrate this breakthrou
929.
by IanCal 2y ago
Is it preferable to have none or enough?
930.
by IanCal 2y ago
Not in the example given there.

More ›