Simple question, difficult solution. I can’t work it out. I have a server at home with a site-to-site VPN to a server in the cloud. The server in the cloud has a public IP....
Short answer: Don’t bother, it’s too complex to setup (unless your app is HTTP or supports the PROXY protocol). You better read your proxy logs instead.
Long answer: What you want is called “IP transparency” and require your proxy to “spoof” the IP address of the client when forwarding packets to the remote server. Some proxies do it (Nginx plus, Avi Vantage, Fortinet) but are paid services. I don’t know for free solutions as I only ever implemented it with those listed above.
This require a fairly complex setup though:
0. IP address spoofing
The proxy must rewrite all downstream request to spoof the client IP address, making it look like the traffic originates from the client at the TCP layer.
1. Backend server routing
As the packet will most likely originate from random IP on the internet, your backend server must have a way to route back the traffic to the proxy, instead of it’s default gateway. Otherwise you’d implement what is called "Direct Server Return*, which won’t work in your case (packet will be dropped by the client as originating from your backend server directly, and not from the proxy).
You have two solutions here:
set your default gateway to the proxy over its VPN interface (don’t do that unless you truly understand all the implications of such a setup)
use packet tagging and VRF on the backend server to route back all traffic coming from the VPN, back to the VPN interface (I’m not even sure this would work with an IPsec VPN though because of ACL…)
3. Intercept and route back return traffic
The proxy must be aware that it must intercept this traffic targeted at the destination IP of the client as part of a proxied request. This require a proxy that can bind on an IP that is not configured on the system.
So yeah, don’t do that unless you NEED to do that (trust me as I had to do it, and hated setting it up).
Edit: apparently haproxy supports this feature, which they call transparent mode
A while ago someone helped me to eliminate the top tab bar so i could just have my Tree Style Tab on the side. It works great but i was wondering if there was any way to make that a toggle? Cause sometimes I move my computer to a smaller monitor and the horizontal space that it takes can be inconvinient.
Here's the snippet for removing it in the first place:
/* hides the native tabs */ #TabsToolbar { visibility: collapse; }/* leaves space for the window buttons */ #nav-bar { margin-top: -30px; margin-right: 0px; margin-bottom: -4px; }
Also, since i have you here, i couldn´t manage to create a floating o just an existing "three buttons" (like the minimize maximize and close buttons). So if you know about any one of those it´d be really helpful.
Recently my NAS took some physical damage and the HDDs are not too happy about it. Most of my video files are partially corrupted. Meaning, they report some errors when checked with ffmpeg[^1], and when you watch them they’ll sometimes freeze or skip a few seconds, but they’re not so corrupt they won’t play. So, the vast...
I got my car (2020 Ford Fusion Hybrid SE) new 3 years ago at $25k for a 6 year loan @ 0% interest for entirety of loan, $350 a month payment. I’m about halfway paid off and have about $12.5k left on it. What should I do? I just get sick of paying $350 a month.
If you’re paying 0% it won’t hurt to keep paying it monthly. If you have 12.5k laying around then invest it and make some money. Obviously if you had anything above 0% interest on the loan then paying it off would never be a bad thing.
Is it absolutely 0% or is it 0% with a $10/month administration fee? If the former, don’t pay it off early, just set up a standing order/direct debit and let it pay itself down. If the latter, you need to calculate the comparison rate (which will get higher the closer you get to zero balance), and work out what the break even is. Then carry on paying until you hit the point that the effective interest is greater than the interest on your savings account and at that point pay it off in full.
Setting aside the talk about interest, etc., I would double check the terms of your loan.
Almost no lending institution is going to give you $25k at 0% for the life of the loan. They wouldn’t be making any money. In fact, servicing the loan (taking payments, etc.) would cost them money.
The days of the 0% for X months financing are coming rapidly to an end as inflation and the federal reserve interest rate both go up. But even those loans, while they incentivize the buyer to pay the loan off early, will still apply interesting – sometimes, if you’re not careful, all the deferred interest for the past X months, which is extra shitty – after the 0% period is complete. I think the only major manufacturer this season that’s offering any sort of 0% deal is NIssan, and they’re sort of going all in, probably in an attempt to move excess inventory.
It’s just not a thing to see a lender not charge interest. It’s like, going to Taco Bell and ordering a burrito for $0.
There are, of course, some exceptions. The dealer may be subsidizing the interest payments. The lender/servicer may be pulling some (probably) illegal shit and calling “interest” a “service fee” or are perhaps charging a “payment processing convenience fee” to make their money. The dealer / automaker may be paying the servicer’s bills on the backend, but don’t think for a moment that cost isn’t baked into the sale price of the vehicle. I just can’t see a case where a lender would be okay working for free.
I’m stoked that they’re making this! Please buy it! There isn’t anything else in the same class as the Studio Display (5k/27”) to my knowledge which was why Apple had to make the Studio Display, but also why they can charge so much for it! It’s nice that it works with Windows for people who use both macOS and Windows, but I’m guessing exactly 0 people buy this exclusively for their XPS/HP laptops.
I think the introductory pricing will come down within a few months. I stated in another comment, but maybe ~$1000 by holiday/ Black Friday season? This looks like a good product, but it’s hard to imagine why someone would spend the same amount for a similar product that isn’t first party (Apple), even if this has some nice included features (matte display and adjustable stand) that Apple charges for. I think long term, the best marketing for this product is “it’s like the Apple display, but costs way less.”
Vielleicht versteht einer von uns die Auswertung falsch, für mich bedeutet das “in 0% der Meetings gab es eine hohe Teilnehmerquote”. Was auch immer die Teilnehmerquote in dem Fall ist.
Das andere bezieht sich ja auch darauf, in wie viel % der Meetings das so war.
Andere Frage: wo findet man das? Mich würde interessieren ob ich bei “kein Multitasking” über 0% bin.
By the way. Another way is to not boot to run level 5 / graphical but tell systemd to only to say level 3. If your in 5 you can also tell it to go to say 3. I cannot remember the commands but maybe someone else does.
If you did not know Linux boot system defines 6 run levels 0…5. You can boot to any of them. 0 is shutdown, 1 is emergency/single user, 3 is usually non-graphical mode, 5 is usually graphical. How you use them depends on your Linux distribution and if it uses systemd.
Because as much as trains and buses are great for everyday commuter movement (and having amenities within walking distance is key as well), there’s two issues:
Changing the infrastructure and zoning of an existing city is much easier said than done. Ripping up concrete, tearing down existing business and homes to increase densification, that’s a huge undertaking.
Trains never replaced the horse drawn carriage. You can never fully eliminate the need for cars because sometimes you need to move something big like a couch. Even if there’s less cars on the road, it’ll never be 0, as this also includes things like ambulances, and fire trucks that can’t rely on schedules.
I’ve got two contentious ones (for different reasons), but I stand by them. They hold 2 of my 3 top all-time "played’ positions despite having gotten them recently in life (the third is Dominion, which misses this list by just a few centimeters)
Game 1? Spirit Island. I think it’s the best cooperative survival strategy ever made. The “almost” in its perfection is the need to ease a new player in, but my experience is that literally anyone can “click” with a low-complexity spirit and run with it indefinitely. And then the game has room to grow in every direction. Complexity? More/different spirits. Difficulty? The base difficulty leads you to nearly 100% win rate, but adding events or adversaries give a slow (but real) increase up to the highest difficulties that approach a 0% win rate even at the highest level of play. Why is this contentious? It’s popular and new, and everyone (including me) is hesitant to pick a game like that as a GOAT.
Now the opposite side of the coin, possibly edging out Spirit Island slightly… Kingdom Death: Monster. Let’s get the “not quite perfect” out of the way fast - the price is absolutely oppressive, and buying my used copy was as much of a risk as an investment. Knowing the current expansion list is over a grand and there’s another expansion set sitting in the store at $1250, that’s a hard pill to swallow if you need to own everything about a game! But you talked theme in Terraforming Mars - KD:M is the king of theme. It’s a brutalist masterpiece that has no hesitation turning a mistake into a “time to start from scratch” campaign-wipe at the 30 or 40 hour mark. Its mechanics are reasonable enough for any gamer to pick up, but its playstyle is simple enough that even a non-gamer can become fully immersed as long as they have someone running the rules in the background for them.
As someone who has tried both and went back to pihole for no reason other than “why not?” – it works as intended, does everything accordingly and I have 0 issues running it plus 2x unbound dns servers.
Inject liberally into veins switches, connectors, and other electronic moving parts that I’d like to be waterproof. (0. Cover PCB in nail polish or specially-made products)
yeah thats a good example and it shows weird the number 0 is compared to the positive integers. it seems like a lot of the time things are first “defined” for the positive integers and then afterwards the definition is extended to 0 in a “consistent way”. for example, the idea of taking exponents a^n^ makes sense when n is a positive integer, but its not immediately clear how to define a^0^. so, we do some digging and see that a^m+n^ = a^m^a^n^ when m and n are positive integers. this observation makes defining a^0^=1 “consistent” with the definition on positive integers, since it makes a^m+n^ = a^m^a^n^ true when n=0.
i think this sort of thing makes mathematicians think of 0 as a weird index and its why they tend to prefer starting at 1, and then making 0 the index for the “weird” term when it’s included (like the displacement vector in affine space or the constant term in a taylor series).
“Our primary conclusion across all scenarios is that without enough fresh real data in each generation of an autophagous loop, future generative models are doomed to have their quality (precision) or diversity (recall) progressively decrease,” they added. “We term this condition Model Autophagy Disorder (MAD).”...
The "adequate covering" of our distribution p is also pretty self-explanatory: We don't need to see the statement "elephants are big" a thousand times to learn it, but we do need to see it at least once:
Think of the p distribution as e.g. defining a function on the real numbers. We want to learn that function using a finite amount of samples. It now makes sense to place our samples at interesting points (e.g. where the function changes direction), rather than just randomly throwing billions of points against the problem.
That means that even if our estimator is bad (i.e. it can barely distinguish real and fake data), it is still better than just randomly sampling (e.g. you can say "let's generate 100 samples of law, 100 samples of math, 100 samples of XYZ,..." rather than just having a big mush where you hope that everything appears).
That makes a few assumptions: the estimator is better than 0% accurate, the estimator has no statistical bias (e.g. the estimator didn't learn things like "add all sentences that start with an A", since that would shift our distribution), and some other things that are too intricate to explain here.
Importantly: even if your estimator is bad, it is better than not having it. You can also manually tune it towards being a little bit biased, either to reduce variance (e.g. let's filter out all HTML code), or to reduce the impact of certain real-world effects (like that most stuff on the internet is english: you may want to balance that down to get a more multilingual model).
However, you have not note here that these are LANGUAGE MODELS. They are not everything models.
These models don't aim for factual accuracy, nor do they have any way of verifying it: That's simply not the purview of these systems.
People use them as everything models, because empirically there's a lot more true stuff than nonsense in those scrapes and language models have to know something about the world to e.g. solve ambiguity, but these are side-effects of the model's training as a language model.
If you have a model that produces completely realistic (but semantically wrong) language, that's still good data for a language model.
"Good data" for a language model does not have to be "true data", since these models don't care about truth: that's not their objective!
They just complete sentences by predicting the next token, which is independent of factuallity.
There are people working on making these models more factual (same idea: you bias your estimator towards more likely to be true things, like boosting reliable sources such as wikipedia, rather than training on uniformly weighted webscrapes), but to do that you need a lot more overview over your data, for which you need more efficient models, for which you need better distributions, for which you need better estimators (though in that case they would be "factuallity estimators").
In general though the same "better than nothing" sentiment applies: if you have a sampling strategy that is not completely wrong, you can still beat completely random sample models. If your estimator is good, you can substantially beat them (and LLMs are pretty good in almost everything, which means you will get pretty good samples if you just sample according to the probability that the LLM tells you "this data is good")
For actually making sure that the stuff these models produce is true, you need very different systems that actually model facts, rather than just modelling language. Another way is to remove the bottleneck of machine learning models with respect to accuracy (i.e. you build a model that may be bad, but can never give you a wrong answer):
One example would be vector-search engines that, like search engines, retrieve information from a corpus based on the similarity as predicted by a machine learning model. Since you retrieve from a fixed corpus (like wikipedia) the model will never give you wrong information (assuming the corpus is not wrong)! A bad model may just not find the correct e.g. wikipedia entry to present to you.
My Qbittorrent refuses to seed for some reason, I have really good internet speed and wanna help with seeding but it just constantly says 0 upload even though the limiter is set to infinite.
Those are valid points and make some practical sense, but I’ve talked too much with mathematicians about this so let me give you another point of view.
First of all, we do modular arithmetic with integers, not natural numbers, same with all those objects you listed.
On the first point, we are not talking about 0 as a digit but as a number. The main argument against 0 being in N is more a philosophical one. What are we looking at when we study N? What is this set? “The integers starting from 0” seems a bit of a weird definition. Historically, the natural numbers always were the counting numbers, and that doesn’t include 0 because you can’t have 0 apples, so when we talk about N we’re talking about the counting numbers. That’s just the consensus where I’m from, if it’s more practical to include 0 in whatever you are doing, you use N0. Also the axiomatization of N is more natural that way IMO.
BattleBit Remastered - Update 1.8.1: Party Codes, Stat Protection, QoL Updates, Bug Fixes (store.steampowered.com)
cross-posted from: lemmy.world/post/1635457...
double-o rule (lemmy.blahaj.zone)
Transcription: “They call me 007. 0 idea why my game crashes. 0 idea how to configure mods. 7 types of copper ore.”
Proxy to TCP port with real IP
Simple question, difficult solution. I can’t work it out. I have a server at home with a site-to-site VPN to a server in the cloud. The server in the cloud has a public IP....
Repair corruption without full re-download?
Recently my NAS took some physical damage and the HDDs are not too happy about it. Most of my video files are partially corrupted. Meaning, they report some errors when checked with ffmpeg[^1], and when you watch them they’ll sometimes freeze or skip a few seconds, but they’re not so corrupt they won’t play. So, the vast...
I Played Road 96 AND the Prequel -- Yes, They Made a Road 96 Prequel (review, minor spoilers)
What is Road 96?...
Is it worth paying the reaminder of my car loan off?
I got my car (2020 Ford Fusion Hybrid SE) new 3 years ago at $25k for a 6 year loan @ 0% interest for entirety of loan, $350 a month payment. I’m about halfway paid off and have about $12.5k left on it. What should I do? I just get sick of paying $350 a month.
Samsung Debuts New $1,599 ViewFinity S9 5K Display to Compete with Apple's Studio Display (www.macrumors.com)
ich💼💪iel (lemm.ee) German
Is it possible to use exclusively a TTY on Linux?
Hello everyone....
[meme] Trains are 100000x easier to electrify and automate than cars, so why does everyone keep talking about electric and driverless cars? (lemmy.world)
deleted_by_author
Pihole vs AdGuard Home
Hey, not sure if this is the right community, but looking for some information....
Alphabetically sorted months. But WHY? (lemmy.ml)
What do you use Vaseline for?
I have a tub of Vaseline and have hardly scratched the surface. I’m curious whether anyone uses it for anything other than their lips.
Random internet people explaining math better then math teacher (i.imgur.com)
AI model output quality decreases when trained with AI models (futurism.com)
“Our primary conclusion across all scenarios is that without enough fresh real data in each generation of an autophagous loop, future generative models are doomed to have their quality (precision) or diversity (recall) progressively decrease,” they added. “We term this condition Model Autophagy Disorder (MAD).”...
Piracy Wiki + Megathread (lemmy.dbzer0.com)
https://a.imagem.app/bymZLb.png...
Lua (lemmy.ml)