Lucene and ES implement a shortcut for filters that are restrictive enough. Since it's already optimized for figuring out if something falls into your filter set, you first determine the size of that. You traverse the HNSW normally, then if you have traversed more nodes than your filter set's cardinality, you just switch to brute forcing your filter set distance comparisons. So worst case scenario is you do 2x your filter set size vector distance operations. Quite neat.
Oh that's nice! Any references on this shortcut? How do you activate that behavior? I was playing around with ES, but the only suggestion I found was to use `count` on filters before deciding (manually) which path to take.
You know, humans pick up hobbies like cycling or running which they do consistently for years, listen to specific music genres or even electronic music which is mostly just a beat, hang up a painting they like in the living room and look at it for years and years, go out to their favorite place to eat consistently or cook the same passed down family recipe, and in so many other aspects avoid sudden changes, and you're surprised that for video games we enjoy the same formula repeatedly?
If you walked in to my living room and saw 17 almost identical paintings where maybe one is styled to look like papercraft, one has a little dinosaur, one has a racoon tail ect you'd rightly think i was a bit mad even before i announced i'd paid $80 for each one.
Now if i had 17 unique paintings exploring a variety of motifs and styles, each one with a story to tell that would actually be worth talking about.
Actually, I would think the artist is mad for selling those paintings so cheap. Paintings usually command a much higher price tag.
Then I’d compliment your ability to create a flowing theme throughout the house.
The problem with your analogy here is that art for the home (or anywhere outside a museum or gallery) is generally bought to compliment the overall aesthetics of the building, rather than to be enjoyed in isolation.
> Now if i had 17 unique paintings exploring a variety of motifs and styles, each one with a story to tell that would actually be worth talking about.
That’s called “eclectic”, which is basically an artsy way of saying “mismatched”. Some people dig that style. Personally I don’t.
Likening a game to a painting is just a false premise. Games are a unique medium that can in themselves hold lots of different things which inherently makes them hard to compare wholesale to other mediums. They can be similar to Chess in the sense of requiring strategy or a physical sport in that they can be almost entirely composed of the skill expression of the players. They can even hold paintings or novels in their entirety, or do something entirely unique that just can't be done in other mediums (my favorite example of that is Outer Wilds).
Why are you so judgmental of what kinds of paintings people hang on their walls? Just hang the paintings on your walls that you like and leave everybody else alone.
I’m assuming that’s the fully burdened rate, i.e., salary, benefits, taxes, overhead, profit, etc that employees never see, even though they should.
I remember getting a sheet from an employer early in my career that fully broke down the cost of benefits and taxes and showed me the full cost of just my employment, not including overhead, profit, etc. it was rather eye opening because although I kid of knew it from accounting and finance, it never really impacted me quite as much before seeing the numbers.
If the system didn't structure it that way, everyone would know the numbers and protest. As it is people pay up for the most part, they even defend the concept of taxing, supposedly going to public services.
>I'm still looking for a systematic approach to make a hybrid search (combined full-text with embedding vectors).
Start off with ES or Vespa, probably. ES is not hard at all to get started with, IMO.
Try RRF - see how far that gets you for your use case. If it's not where you want to be, time to get thinking about what you're trying to do. Maybe a score multiplication gets you where you want to be - you can do it in Vespa I think, but you have to hack around the inability to express exactly that in ES.
I don't know how this is something related to AI - you could polish and embellish your resume before LLMs too, I'm fairly sure. I guess this gets the clicks.
Not being to remember small details about certain projects is also perfectly fine for people who have worked for more than a couple of years. Unless you can discover a pattern of lying like the author supposedly did then I would just be perfectly fine moving on to another topic.
Agreed, the “I used AI” part is just the 2025 version of “I did my research on your company and then lied about my experience to make me sound like a better fit”.
The twist on “I used AI” to this though is that everyone comes out looking the same. They all have the same resume format, made by the same tool, stuffed with the same keywords.
Fair for the ones who don’t put in any effort, but I don’t buy this generalization for the folks who are real people in the middle between “completely unqualified” and “telling the truth about their experience”.
Any effective screening strategy is going to catch the liars who do it only a little with some probability.
Not least because being willing to be dishonest during an interview is a strong signal the candidate will be dishonest while they are employed as well, and companies want very much to not hire those people.
My memory had held back my career I am sure. I can't regurgitate the minute details of impact I did even 12 months ago, just broad strokes... so I prep as best I can but it probably sounds like I am lying. Now with AI and everyone is suspicious it is worse. Got downleveled to 4yoe level yoe from where I am 20yoe but I needed a job so.
I have what probably qualifies as the relatively-recently-named "severely deficient autobiographical memory".
Notes, notes, notes. Then review them before an interview. Not bullet-point notes of things that happened (that's fine too, but not just that) but make stories when they're very fresh, like, right after they happen. You won't be able to turn raw bullet points into a story later, you'll forget too much.
Then take some time to match stories to common interview questions. That's your prep document. Feel absolutely free to fill in gaps where needed, most folks' "real" memories of these things are half wrong anyway, and there may be times you literally couldn't have an acceptable answer to a common question without making some of it up, because you didn't take useful-enough notes. What are you going to do, fail every interview that asks that question forever? No, just make the story you need, connect it to reality as much as possible, and move on. But do it ahead of time. And you only need to do this once per such question. Perhaps you'll even manage to take notes on a less-invented story later (I've found that nearly all of these stories need a little invention, though, even if you have perfect notes, to fit into the acceptable range of responses)
It’s called a career document or a brag document. I update mine every quarter. It’s a detailed summary of the projects I worked on in STAR format including challenges I faced.
I think in this case the candidate didn’t even know enough to embellish the resume unassisted. Their nonsense response on rate limiting showed that they had no idea why you would rate limit or under what circumstances. Ditto for paginating data.
AI allowed them to add plausible work to their resume that they couldn’t have come up with on their own.
4.8b words on English Wikipedia. Knowledge cutoff of 6 months. A valid use case is to search across Wikipedia and ground your
answers.
Trivially proves that RAG is still needed.
I was using Firefox exclusively for years, but when I sold my Macbook and bought a Thinkpad and installed Linux on it, I grew pretty annoyed by Firefox.
Specifically, I couldn't view my 360 videos or photos on Google Images or Immich at anywhere near acceptable performance. The videos, recorded at 30fps, would get maybe 5fps. This was weird, because I have a fairly beefy laptop, it should be able to handle these videos just fine (especially since my iPhone handled it just fine).
After a bit of debugging, it appears that there's a bug in how it's writing for the shader cache, and as such there was no hardware acceleration. I found a bug filed about my issue [1], and I didn't really feel like trying to fix it, because I didn't want to mess with Mesa drivers. I just installed Chromium and that's what I'm using right now, and it worked with my 360 videos and photos absolutely fine.
I want Firefox to succeed, but that really left a bad taste in my mouth; it's not like it's weird to want my browser to be hardware accelerated.
How big is your monitor? I can only see about 10-15 tabs on my 4k monitor before Firefox starts scrolling them off the screen. I regularly have 2-3x that on Chrome before tabs stop showing up.
I have 54 tabs open right now. The Sideberry extension lets you view them in the left sidebar. They're nested so that collapsing a root tab will also collapse all child tabs. There are also super tabs (Sideberry calls them "Tabs panels") so you can switch between entire groups of tabs.
1,740 tabs open right now on my wife's Firefox and it seems to be operating just fine. Sounds like something's wrong with your Firefox. I recommend a refresh which can be found under about:support
10, 20, even 30 i can understand. More is the equivalent of forgetting to empty the kitchen trash can and still filling it until the smell is horrible.
someone got to tell her there is a cross on the right to close the tab.
I used to bookmark everything into Diigo. Then their Firefox extension stopped working... and I haven't got a cross-platform, cross-browser process up and running again.
Is there a tab session manager that does that, and lets me send tabs from my current session to another session? E.g. I'm on my "Writing C for a hobby" session and quickly search for something cooking related, and then need to send that to my cooking session?
757 is too many though. If nothing else, if you couldn't find an extension that works, you could just use different browser profiles, or save the links in text files. It's just so ridiculous to have that many open when, without any doubt, you don't need them open all at once.
I use tab session manager on firefox. It doesn't easily let me shift around tabs inside a session, if I want to combine sessions I have to open both and save as a new session. It does allow duplicating and trimming tabs from a session though.
If you need better session management capability, you could probably get an LLM to extend/fork an existing extension to add what you need with about 30 minutes work.
Firefox works great with dozens of open tabs. The only thing Chrome has going for it is tab groups. Firefox has Tab Style Tree, which is a decent substitute.
You don't need an extension. Right-click on a tab and "Select All Tabs". Right-click again and it has the option "Close 1,122 Tabs". Your number may be smaller.
That only works if you've got a single window open. For myself, I keep ~10+ windows open, with then ~8 tabs per window. Note this is only practical on a tiling window manager. Anyway, the tab count extension may still be the way to go.
You can't see all 50-70 tabs on a normal 27" monitor; Chrome will squish them almost indefinitely, and Firefox forces a large minimum tab width that makes the tab bar scroll forever and then you forget half the tabs you have going and everything's bad. I tried to switch and stopped because of this. I'll hang on until ubo really stops working, I guess, and then try to figure something else out.
People wait for things all the time and if it can take away the hassle or driving your kids somewhere, maintaining multiple vehicles for multiple family members etc. I'd bet that it does take off. Public transport doesn't solve groceries or other types of shopping though even in highly connected cities, especially for people with health problems or just aging ones.
Imagine a parent today driving to work and dropping off their two kids at school on the way. Instead of one car making three stops, you're suggesting three robo taxis would accomplish the same thing. What do you think already clogged roads would look like in this future? The only scalable solution is public transit, walking, and cycling.
I used to live in Germany, which had lots of excellent public transit, walking, and cycling. There was still quite a bit of car traffic. If even Germany's public transport isn't enough to eliminate cars, then we should put some effort into improving car transport, in addition to whatever we do with buses and bike lanes.
To that end, I'm not convinced automation will make things worse.
If the kid's school is near the parent's route to work, there's nothing stopping the parent from saving money by taking the kid in the same taxi.
If the kid's school is in the opposite direction, then a separate robotaxi can be more efficient. What matters is total system mileage. If a robotaxi takes the kid, then picks up a commuter starting near the school, then we save the parent's trip from the school back to their starting point.
