Seems like a great deal.. 😂😂

Seen on a lottery commercial:

Guy 1: That can't be right check it again.
Guy 2: Computers can't be wrong, they're programmed that way!

I wrote an app that tells me if a lottery ticket is winning. It takes a picture of the ticket, does OCR, finds the number lines and compares them with a remote json.
I live next door to a lottery shop.

Holy shit balls it works (so far)

-
So I’ve been working on a project... well feature for the past year (yes 12 whole fucking months - anti agile I know)
And today I got to merge that bitch of a pull request into the current working branch and deploy it to UAT - no conflicts 😵I think I need to put the lottery on tonight!
And some how, by some stupid lucky roll of the dice it just works.

I have never felt so afraid and delighted in my life!

This happend to me around 2 weeks ago. For some reason, I decied to post this now.

I won the lottery, yey! I mean, bot really, but I am <19yo student, "less than junior dev" in my office, but sonce I am the only one who is capable of working with hardware, I was working month back as a sysadmin for a few days. Our last sysadmin was really good working but really, really toxic guy, so he got fired on a spot after argument with some manager or whatever, no big deal, we could have another guy hired in a week. But, our backup server literally was on fire, all data probably dead because bad capacitor or whatever. This was our only backup of everything at the time. Everyone in full fucking panic mode, we had literally no other working HW we could use for backup, but then comes me, intern employed on his first dev job for 3 months. That day I bought some HW for my own personal server at home (Intel NUC with some Celeron, 4GB DDR4 RAM and two 240GB SSDs for RAID 1. My manager asked everyone in the office for sollution how to survive next 4 days before new server arrives. People there had no idea what tk do and no knowedgle about HW, I just came from a break and offered my components for a week, since there was noone else who can work with HW, servers and stuff like this, manager offered me $500+HW cost if I, random intern, can make it work. I installed Debian on that little PC, created RAID1 from both SSDs, installed MySQL server and mirrored GIT server from our last standing server (we had two before one of them went lit 🔥), made simple Python script to copy all data on that RAID, with some help of our database guy copied whole DB from production to this little computer and edited some PHP so every SQL request made on our server will run on that NUC too. Everything after ±2 hours worked perfectly. Untill a fucking PSU burned in our server and took RAID controller with him in sillicon heaven next night, so we could not access any data unltill we got a new one. Thanks to every god out there, I was able to create software RAID from survived HDDs on our production server and copy all data from that NUC on the servers software RAID and make it working at 3 AM in the night before an exam 😂. Without this, we would be next ±40 hours without aerver running and we might loose soke of our data and customers. So my little skill with Linux, Python, MySQL and most importantly my NUC hardware I got that day running as a backup server saved maybe whole company 😂.
Btw, guess who is now employee of the year with $2500 bonus? 😀
Sorry for bragging and log post, but I was so lucky an so happy when everything worked out, good luck to all sysadmins out there! 👍

TL:DR: Random intern saved company and made some money 😂

!rant

Today I bring happy news. First company I interviewed at clicked so well, both personally and technically, and they expressed an eagerness to hire me on the spot. I figured we might as well talk salary to compare them against other interviews. The offer they made me was so good I decided to sign there and then. They said they participate in a fair wage program but for me this is absolutely the dream. I get lots of nice perks to boot. And they've already mailed me some tech documentation to go over so I can prepare, as I'll be working with the latest front end stuff and of course my trusty .NET (and yes I asked it'll be C#, haha).

I can't even begin to express how great this is. The last decade I've been unemployed for several years in total, and vastly underpayed when I was employed. I've worked in some toxic environments, been falling behind on tech and wrote a lot of rubbish code as a result of that. But it seems that somehow all the hard work I did put in paid off by taking a chance when it presented itself and go in accepting I might fail horribly. And I did bomb the tech questions actually. But they let me explain myself and come to answers together and saw beyond the black and white.

In short I feel like I've won the work lottery and will start 2018 in style. Part of me is still scared though, that there will be a mistake or a catch or even somehow I'll ruin everything. But that is the risk in life and I'm just going to have to deal. What I can control and will do is my very best, because I want to keep succeeding and have a great future career. And I hope I can inspire others in the same boat with my actions too.

I've found an affordable flat and it feels like winning the lottery. Before this one I wasn't even getting the overpriced ones!

*discards the tent*

I've recently received another invitation to Google's Foobar challenges.

A while ago someone here on devRant (which I believe works at Google, and whose support I deeply appreciate) sent me a couple of links to it too. Unfortunately back then I didn't take the time to learn the programming languages (Python or Java) that Google requires for these challenges. This time I'm putting everything on Python, as it's the easiest language to learn when coming from Bash.

But at the end of the day.. I am a sysadmin, not a developer. I don't know a single thing about either of these languages. Yet I can't take these challenges as the sysadmin I am. Instead, I have to learn a new language which chances are I'll never need again outside of some HR dickhead's interview with lateral thinking questions and whiteboard programming, probably prohibited from using Google search like every sane programmer and/or sysadmin would for practical challenges that actually occur in real life.

I don't want to do that. Google is a once in a lifetime opportunity, I get that. Many people would probably even steal that foobar link from me if they could. But I don't think that for me it's the right thing to do. Google has made a serious difference by actually challenging developers with practical scenarios, and that's vastly superior to whatever a HR person at any other company could cobble together for an interview. But there's one thing that they don't seem to realize. A company like Google consists of more than just developers. Not only that, it probably consists - even within their developer circles - of more than just Python and Java developers. If any company would know about languages that are more optimized such as C, it would be Google that has to leverage this performance in order to be able to deliver their services.

I'll be frank here. Foobar has its own issues that I don't like. But if Google were a nice company, I'd go for it all the way nonetheless - after all, they are arguably the single biggest tech company in the world, and the tech industry itself is one of the biggest ones in the world nowadays. It's safe to say that there's likely no opportunity like working at Google. But I don't think it's the right thing. Even if I did know Python or Java... Even if I did. I don't like Google's business decisions.

I've recently flashed my OnePlus 6T with LineageOS. It's now completely Google-free, except for a stock Yalp account (that I'm too afraid to replace with my actual Google account because oh dear, third-party app stores, oh dear that could damage our business and has to be made highly illegal!1!). My contacts on that phone are are all gone. They're all stored on a Google server somewhere (except for some like @linuxxx' that I consciously stored on device storage and thus lost a while back), waiting for me to log back in and sync them back. I've never asked for this. If Google explicitly told me that they'd sync all my contacts to my Google account and offer feasible alternatives, I'd probably given more priority to building a CalDAV and CardDAV server of my own. Because I do have the skills and desire to maintain that myself. I don't want Google to do this for me.

Move fast and break things. I've even got a special Termux script on my home screen, aptly named Unfuck-Google-Play. Every other day I have to use it. Google Search. When I open it on my Nexus 6P, which was Google's foray into hardware and in which they failed quite spectacularly - I've even almost bent and killed it tonight, after cursing at that piece of shit every goddamn day - the Google app opens, I type some text into it.. and then it just jumps back to the beginning of whatever I was typing. A preloader of sorts. The app is a fucking web page parser, or heck probably even just an API parser. How does that in any way justify such shitty preloaders? How does that in any way justify such crappy performance on anything but the most recent flagships? I could go on about this all day... I used to run modern Linux on a 15 year old laptop, smoothly. So don't you Google tell me that a - probably trillion dollar - company can't do that shit right. When there's (commercialized) community projects like DuckDuckGo that do things a million times better than you do - yet they can't compete with you due to your shit being preloaded on every phone and tablet and impossible to remove without rooting - that you Google can't do that and a lot more. You've got fucking Google Assistant for fucks sake! Yet you can't make a decent search app - the goddamn thing that your company started with in the first place!?

I'm sorry. I'd love to work at Google and taste the diversity that this company has to offer. But there's *a lot* wrong with it at the business end too. That is something that - in that state - I don't think I want to contribute to, despite it being pretty much a lottery ticket that I've been fortunate enough to draw twice.

Maybe I should just start my own company.

With luck like this I'll be buying a lottery ticket next.

I wrote my first line of code at 12. I fell in love with it and continued. I'm 25 now and I'm a software engineer. I don't even have time or energy to work on personal projects anymore. Writing code isn't a hobby anymore. It's a means to survive. Why/how did this happen? When will building things be fun again? Before landing my first job as an engineer, not once did I consider salaries, equity, atmosphere, nor any of the other amenities (or lack thereof) of code as a profession. But, I don't even know when any of that fell into the picture and they've managed to suck the novelty out of a really cool pastime. I'm essentially a well-paid robot. Who did this? What's happening? What can I do to find the freedom I once had? When did I become just another cog in a machine? Should I try my hand in business, bent on making a lot of money so I can retire early and have time to experiment again? Is that unrealistic? Should I buy lottery tickets every paycheck? We only get one life and I realized this. I'm panicking because I know I'm not enjoying myself and that I'm not on track to leave the world better than it was when I was born into it. So much loss. I'm grateful, but this is not cool at all. I want my hobby back.

"Can you zoom in to this teeny tiny spot on this photo and sharpen it like how they do it in CSI?"
OR
"You can uncensor this, yeah?"

Yeah let me poop out the winning lottery numbers while I poop out nonexistent pixels for you.

This day I have received the most glorious news in e-pistolary form. For some years, I was suffering in support of a client who was, well, insufferable. My presence there paralleled the divine comedy in both essence and fact.

I opened the missive, expecting another plea to bail them out of whatever clusterfuck they found themselves in. Instead, what I found was something truly magical.

"Hey Human,

I hope this finds you well. I'm not sure if you remember a few years back, we were trying to decide between IBM Cloud and AWS. Well, after years of battling FF*, we're finally moving ahead with AWS. He failed one too many times to deliver anything visibly. After you left, there was no one left he could use to steal credit, ideas, and work.

FF is still pushing to have them use IBM cloud as a "warm backup" in the event "AWS fails." We will see where that goes.

I figured you'd like to know; you were the void in the wilderness for a long time. I don't want to think about how much time we could have saved if we had just listened.

PeeEm**"

This event represents a personal victory, albeit belated, over a few peoples' absurd amount of privilege. Towards the end, I was vicious about my contestation to the insanity of adopting a desperate hedge attempt-as-cloud offering from a failing company. Some examples:

// cloud 'strategy meeting'
Moi: What cloud platform are we looking at using?
FF: We're looking at IBM cloud and AWS as a second.
Moi: Why is that? I understand you're obligated to rep your offering first, but that decision doesn't seem to have the customer's best interest at heart.
FF: IBM cloud is a market leader; AWS isn't as good.
Moi: I see. I mean, that's the tech equivalent of the company's fleet management considering monkeys on tricycles as a strong competitor to service trucks, but I get what you mean.

// steering meeting
Director: Who can we look to as an example? Who is currently using the IBM cloud?
Moi: No one; they account for a single-digit portion of the actual cloud market. Their long game to sell you a "Hybrid Cloud," which means put some front end payload in a CDN, and buy n-frame units of IBM z servers for the DC with IBM gateway appliances acting as connective tissue. So it's not the cloud at all, really.
Director: How does it compare in cost?
Moi: It's generally 40% more expensive than other clouds, and it only goes higher as you option their software.
Director: What about Watson? I hear Watson is good?
Moi: It's a brand name. Most of the "Watson" product is just a facade on top of FOSS products like Spark, Hadoop, Elasticsearch, etc.
Director: Those were words. They sounded good. FF say it's good tho so we'll believe him because we're from the same city.
Moi: *deletes Director from LinkedIn*

Moral of the story: Never trust a vendor that only recommends their products.

*FF = FatFuck - an embarrassingly rotund individual whose girth is roughly equivalent to his height. He shit his way into an IBM architect position in his mid-20s purely due to winning the visa lottery. He had fake hair glued to his head for his wedding to hide his male pattern baldness; his arrange-married wife undoubtedly cries herself to sleep after sex.

**PeeEm - the then project manager, now portfolio manager of some satellite projects. An overall decent human being, capable.

"Look I won a {}!".format(lottery)

Felt great because I overclocked my 7700k to over 5.2GHz (with a tower fan). A few days later I was streaming Overwatch, won a match, blue screened...
So close to winning the silicone lottery!

Fuck, every time I read something about NFTs I lose another bit of faith in humanity I didn't even know I had.

I mean, I guess someone could make a bunch of cartoon characters (lets say, some sort of "pookey-man") using compatible styles and colour palettes. And I guess you could buy some of those if there is a product that makes trademark recognition a plausible dream (let's say, a videogame and/or anime).

But if the chance of success is akin to the chance of becoming a billionaire solemly from lottery wins, shouldn't the assets be priced accordingly?
Like a couple bucks a pop, at most?

Dropping a cool 30000 quid on a single lotto ticket sounds like the dumbest thing ever.
And yet at least a couple hundred hollow heads did it.
Fuck everything.

Saw a rare moment. Maybe I should buy a lottery ticket? It might be a winner :)

Pretty sure I'm going to win the lottery with these numbers xD

Rough translation:
Here are your lottery numbers.

No Rows Founds

Me and my coworkers are pretty lucky. Your head of development is a developer himself and our CEO listens to developers on advice and actually tries to understand which parts are hard to build and which parts work very well.

I started off in a MNC company as a junior developer. I entered with candy glasses.
I didn't expect to win the lottery. Of getting abuse by superior.

I stayed for a year, at the project. Constantly being belittled by this team lead. It was awful i enter as a fresh grad. All the new tech were so new and scary at that point.

During my time there, i constantly think that developer is not my stuff.

Ultimately i reach the state of burnout. I reached out to the manager and broke down in his office.

I actually told the manager. "I hate coding"

I remember staying up to 4am just complete a piece of program. To be ready to be push to production the next day. My team lead just come screaming at me saying there is bug.

Upon receiving that message via skype. I broke, tears flow down my eyes.

After which i reach a state of burn out. I start to reach out to external parties for help to get me out of there.

Now i am recovered from the burn out. I am curious of the technology that were utilized in that project. I literally face palm. After understanding the technology it isn't so hard after all. I just didn't gear myself up with the tech.

I still do enjoy working on code.

Just got picked in a national “lottery” for mandatory military training for a year. And I moved to another city more than half a year before, got a very good job. Fml. Europe at that

This supposed to be a rant but I have no words to describe in how big fucking mess I am right now while looking for a new workplace...
It’s most unorganized experience that I am part of...
It’s worse then looking for lost key using bare hands in a shit hole....

With last one at least you know what to expect and you have some chances to make some decisions. The hiring otherwise it’s like lottery...

There are no words to describe it.
Fuck...

Let everyone looking for a job right now have my condolences...

In 2015 I sent an email to Google labs describing how pareidolia could be implemented algorithmically.
The basis is that a noise function put through a discriminator, could be used to train a generative function.
And now we have transformers.

I also told them if they looked back at the research they would very likely discover that dendrites were analog hubs, not just individual switches. Thats turned out to be true to.

I wrote to them in an email as far back as 2009 that attention was an under-researched topic. In 2017 someone finally got around to writing "attention is all you need."

I wrote that there were very likely basic correlates in the human brain for things like numbers, and simple concepts like color, shape, and basic relationships, that the brain used to bootstrap learning. We found out years later based on research, that this is the case.

I wrote almost a decade ago that personality systems were a means that genes could use to value-seek for efficient behaviors in unknowable environments, a form of adaption. We later found out that is probably true as well.

I came up with the "winning lottery ticket" hypothesis back in 2011, for why certain subgraphs of networks seemed to naturally learn faster than others. I didn't call it that though, it was just a question that arose because of all the "architecture thrashing" I saw in the research, why there were apparent large or marginal gains in slightly different architectures, when we had an explosion of different approaches. It seemed to me the most important difference between countless architectures, was initialization.

This thinking flowed naturally from some ideas about network sparsity (namely that it made no sense that networks should be fully connected, and we could probably train networks by intentionally dropping connections).

All the way back in 2007 I thought this was comparable to masking inputs in training, or a bottleneck architecture, though I didn't think to put an encoder and decoder back to back.

Nevertheless it goes to show, if you follow research real closely, how much low hanging fruit is actually out there to be discovered and worked on.

And to this day, google never fucking once got back to me.

I wonder if anyone ever actually read those emails...

Wait till they figure out "attention is all you need" isn't actually all you need.

p.s. something I read recently got me thinking. Decoders can also be viewed as resolving a manifold closer to an ideal form for some joint distribution. Think of it like your data as points on a balloon (the output of the bottleneck), and decoding as the process of expanding the balloon. In absolute terms, as the balloon expands, your points grow apart, but as long as the datapoints are not uniformly distributed, then *some* points will grow closer together *relatively* even as the surface expands and pushes points apart in the absolute.

In other words, for some symmetry, the encoder and bottleneck introduces an isotropy, and this step also happens to tease out anisotropy, information that was missed or produced by the encoder, which is distortions introduced by the architecture/approach, features of the data that got passed on through the bottleneck, or essentially hidden features.

you know what im tires of?

Finding a good domain name for a potential business, unregistered, and then using algorithms, the registrar itself snipes it and cybersquats it as "premium".

In otherwords, if you do find a good name, theres no point becauss it'll just be immediately labelled "premium" by an algorithm and lock you out with 5,000 dollar pay wall.

people in 2003 didnt have to deal with this shit. Registrars should be allowed to do this.

Five domain names now, out of a couple dozen I tried, the five good ones I came up with, all five, "premium".

It wasnt like they were even .coms or common words either. Hell one of them had a number in it.

Nope "we have determined spontaneously, through algorithm, you haves selected what may be a valuable domain name, thank you for the service of identifying it for us, we will now reserve it, even though no one else wants it, at a prohibitively high cost."

Like a homeless women finding a winning lottery ticket in a parking lot, and the rich fucking owner running out demanding that she give him it because it was lost in HIS public parking lot.

Like you motherfuckers dont already have enough? You know what a good domain is? Its a basis for credbility. Its the difference between whether people use your service or not. Its the foundation for excitement or interest.

And here we have this "algorithmically marked as premium" bullshit, fucking the poors out of any chance of even a good start.

"Haahahaha cocksuckers, you're not internet startups in the early two thousands! If you dont habe five grand go drop on a dpmain name that isnt even fucking owned, enjoy staying part of the fucking lowerclass!"

These fuckers. Cant believe this bullshit.

Just another day in motherfucking america, where you have to start rich to even get ahead. just one more way gen x, gen y, and gen z got fuckity fucked right in the ass.

fuck this country so much. fuck it all.

never even gonna have a chance to own a home or anything else.

nobody ever offered me a real fucking chance, not once in my god damned life. not even my fucking parents.

might as well drink myself into a coma.

I got my job by applying on Indeed. I know! Crazy, right? You apply for it and you get it. Pretty much never happens with these job sites because of crappy recruiters and clueless HR drones who tend to post and forget. I must’ve hit the luck lottery that day.

A client wanted his site to be finished earlier. The boss said ok if the client would pay more, and he did. However, a FE dev got sick (virus infection he said) and would have to rest for an extended period of time. The PM came in yesterday and said, "Alex (me), we are running out of time so you will have to take (FE dev's name)'s position because you said that you learned basic html/css when you were in high school." The PM left and close the door before I can turn my head away from the screen and say something (I was like WTF?). One of my BE colleague, who had been asked to do similar sh**t before when a FE guy got sick, pat on my shoulder and said, "Alex, you have won the lottery this time." Q.A.Q..I have just finished a navbar with 4 tabs using Uikit, it took me 90 min already.

Off work at 5:30. In about 10 minutes I'm winning the lottery, then I'm quitting my job and buying the least technical business ever... A brothel.

This is gonna be a long post, and inevitably DR will mutilate my line breaks, so bear with me.
Also I cut out a bunch because the length was overlimit, so I'll post the second half later.

I'm annoyed because it appears the current stablediffusion trend has thrown the baby out with the bath water. I'll explain that in a moment.

As you all know I like to make extraordinary claims with little proof, sometimes
for shits and giggles, and sometimes because I'm just delusional apparently.

One of my legit 'claims to fame' is, on the theoretical level, I predicted
most of the developments in AI over the last 10+ years, down to key insights.
I've never had the math background for it, but I understood the ideas I
was working with at a conceptual level. Part of this flowed from powering
through literal (god I hate that word) hundreds of research papers a year, because I'm an obsessive like that. And I had to power through them, because
a lot of the technical low-level details were beyond my reach, but architecturally
I started to see a lot of patterns, and begin to grasp the general thrust
of where research and development *needed* to go.

In any case, I'm looking at stablediffusion and what occurs to me is that we've almost entirely thrown out GANs. As some or most of you may know, a GAN is
where networks compete, one to generate outputs that look real, another
to discern which is real, and by the process of competition, improve the ability
to generate a convincing fake, and to discern one. Imagine a self-sharpening knife and you get the idea.

Well, when we went to the diffusion method, upscaling noise (essentially a form of controlled pareidolia using autoencoders over seq2seq models) we threw out
GANs.

We also threw out online learning. The models only grow on the backend.
This doesn't help anyone but those corporations that have massive funding
to create and train models. They get to decide how the models 'think', what their
biases are, and what topics or subjects they cover. This is no good long run,
but thats more of an ideological argument. Thats not the real problem.
The problem is they've once again gimped the research, chosen a suboptimal
trap for the direction of development.

What interested me early on in the lottery ticket theory was the implications.
The lottery ticket theory says that, part of the reason *some* RANDOM initializations of a network train/predict better than others, is essentially
down to a small pool of subgraphs that happened, by pure luck, to chance on
initialization that just so happened to be the right 'lottery numbers' as it were, for training quickly.

The first implication of this, is that the bigger a network therefore, the greater the chance of these lucky subgraphs occurring. Whether the density grows
faster than the density of the 'unlucky' or average subgraphs, is another matter.
From this though, they realized what they could do was search out these subgraphs, and prune many of the worst or average performing neighbor graphs, without meaningful loss in model performance. Essentially they could *shrink down* things like chatGPT and BERT.

The second implication was more sublte and overlooked, and still is.
The existence of lucky subnetworks might suggest nothing additional--In which case the implication is that *any* subnet could *technically*, by transfer learning, be 'lucky' and train fast or be particularly good for some unknown task.

INSTEAD however, what has happened is we haven't really seen that. What this means is actually pretty startling. It has two possible implications, either of which will have significant outcomes on the research sooner or later:
1. there is an 'island' of network size, beyond what we've currently achieved,
where networks that are currently state of the3 art at some things, rapidly converge to state-of-the-art *generalists* in nearly *all* task, regardless of input. What this would look like at first, is a gradual drop off in gains of the current approach, characterized as a potential new "ai winter", or a "limit to the current approach", which wouldn't actually be the limit, but a saddle point in its utility across domains and its intelligence (for some measure and definition of 'intelligence').

Every time I go to get some coffee I end up brewing a new pot because I got the last sip...
Every fucking time!
Maybe I should consider playing the lottery...

Using Java for the first time for a homework assignment in uni, everything ran properly the first time.

Gonna go buy some lottery tickets.

Actually finish a side project would be good or win the lottery.

With unlimited time, I'd put resources into the invention/improvement of a container which can be fed photons and is able to bounce them between mirrors for a long time, like days, and can be released at any time.

With that tech, I would build a delayed choice quantum eraser and set it up so that it produces with many photons an interference pattern or strips pattern by choice, representing a bit of information.

Then i would set up many of those devices in a row so that the results are representing bit strings for arbitrary information.

And I will use this time machine, which can send back information, to win the lottery and other stuff.

LinkedIn, if you have the nerves to send me an email, how about you also have some guts and tell me what that "notification" is about. Not that I care in the first place, but generally whenever I see the notification after logging in, I would think the same every time: That was a waste of time to read, even if it was in the email subject.

Like I care about some recruiter and his PHP shit when I've clearly stated as technology I want to avoid, have been deployed for 2 years and am using the tools I prefer.

I also don't care if I appeared in a search.
"Holy fuck, you appeared in that shitshow of a companies shithead recruiter. You better start celebrating like you won the fucking lottery and put your hips in a submissive position like a good boy"

"Thank you for using linkedIn, have you considered giving us money?" - automated bot of linkedIn stuff every now and then.

Is there some kind of fetish I'm missing that makes this an enjoyable or endurable experience?

They keep training bigger language models (GPT et al). All the resear4chers appear to be doing this as a first step, and then running self-learning. The way they do this is train a smaller network, using the bigger network as a teacher. Another way of doing this is dropping some parameters and nodes and testing the performance of the network to see if the smaller version performs roughly the same, on the theory that there are some initialization and configurations that start out, just by happenstance, to be efficient (like finding a "winning lottery ticket").

My question is why aren't they running these two procedures *during* training and validation?

If [x] is a good initialization or larger network and [y] is a smaller network, then
after each training and validation, we run it against a potential [y]. If the result is acceptable and [y] is a good substitute, y becomes x, and we repeat the entire procedure.

The idea is not to look to optimize mere training and validation loss, but to bootstrap a sort of meta-loss that exists across the whole span of training, amortizing the loss function.

Anyone seen this in the wild yet?

A long time ago, I used AIDE (Android IDE) to write a simple application to output random numbers for the lottery. I test it out and head to the gas station. I pulled out my phone to generate numbers, and received something like 36892789, 3, 78921593, 5.

After hours of staring at my code, I transferred and compiled with Eclipse and the problem disappeared.

I keep seeing news story after news story talking about the progress we’re making in AI and how there are not enough developers to fill the jobs in demand for it. I tossed my alternate career choice of aerospace over two decades ago because there were no interesting career paths in what I was capable of doing (not much because I suck at math). And here I am 22 years deep into a Web Dev career that I feel is going to be completely obsolete well before my retirement age. I either need to do the impossible and get good at advanced math so I can even understand the very basics of AI or I need to win the lottery.

Thanks Google!

I'm on vacation.
A friend asked me if I could work on a freelance web project. I was getting bored of summer vacations so I said yes.
It was a website for online lottery and it was already developed by some freelancers.
Owner wanted more freelancers to revamp design and administration panel.

I looked at the site and knew that I had seen the worst design and code of my life.

Frontend was made of two colors only, black and yellow. Out of both, black was more prominent. Moreover it had nothing related to Js as if it was developed as a challenge to be accomplished without java script.

Admin panel and backend was much worse than that. No security practices and deprecated essential libraries.

The nightmare is about to end as I have inducted a much better design from themeforest for frontend.
Backend is in my homebrew php framework.
(Good luck future freelancers 😆)
I'm positive that next edits will be features additions only and no one will blame my code.

I wonder if anyone has considered building a large language model, trained on consuming and generating token sequences that are themselves the actual weights or matrix values of other large language models?

Run Lora to tune it to find and generate plausible subgraphs for specific tasks (an optimal search for weights that are most likely to be initialized by chance to ideal values, i.e. the winning lottery ticket hypothesis).

The entire thing could even be used to prune existing LLM weights, in a generative-adversarial model.

Shit, theres enough embedding and weight data to train a Meta-LLM from scratch at this point.
The sum total of trillions of parameter in models floating around the internet to be used as training data.

If the models and weights are designed to predict the next token, there shouldn't be anything to prevent another model trained on this sort of distribution, from generating new plausible models.

You could even do task-prompt-to-model-task embeddings by training on the weights for task specific models, do vector searches to mix models, etc, and generate *new* models,
not new new text, not new imagery, but new *models*.

It'd be a model for training/inferring/optimizing/generating other models.

can people stop sending me giveaways and shit

it's gambling. fuck off.

my life is not better by getting this nonsense. always gives me anxiety because it hits me with the math of doom. how is this ethical in any way. every regular non-sus shop does this now. it's pervaded every damned business and it's disgusting. makes me depressed for the world

and ofc everyone's favourite daddy government just lets this happen. gambling is bad if we vilify the person, but otherwise it's fine. hell, then government ITSELF told people they could WIN A MILLION DOLLARS if they got the covid shot here. fucked up. chance to die and bigger chance to maim yourself irreparably forever, maybe you'll literally win the lottery though, see we partnered up with the local lottery company. isn't government so great?!

unrealistic dream :

in my 40s, being cto of a market leading product in its v10 stage, whose v1 i created from scratch. me doing nothing except creating the best remote work hustler culture for the company .

i will be making both : the topmost and bottomost engineers/managers give ppts and talks of product that they created, ll system and hl system designs and the decisions that went into it

i also want some powerful management/ ceos as friends, that drive our boat to greater heights and generate tremendous revenue+ fundings + profits

----

realistic dream :

to keep being SE1 (or at worst SE2/tech lead) in a company, do 4-5 hr non impactful work per day and earn 3x the inflation as am doing now. plus somehow get a lottery or something once in a lifetime , that is worth 100x my current income so that i could build a home and 2 cars and a children fund

dev is not a great thing to be dreamt about, but it certainly paves way for a happier and healthier lifestyle if done right

Why isn't Gooogle buying Atlassian to stock up G Suite with Wiki and Slack, then piss off Microsoft and win the Market over with better products?

Im reading everywhere now that MS Teams has the hugest Userbase and so much features to come bla bla bla

Fuck MS Teams, it's shit, looks like it and it's software so I can't smell it. Thank god for that, else it would smell like the afterback of a diaarhetic horse.

Every fucking Tecnician at my company is arguing, that MS Teams is better, because more users are active. WTF Poopface, they have more user, because it comes included with O365???

Our people are so stupid, I bet they won the IT Certification in a Lottery or flew to Turkey to buy it at a Bazaar.

And who the fuck are the Product Managers at Google, gonna hit them a couple of times with a broom to wake up. You fucktards are missing a huge market.

Anyone's got a CKA certificate (Certified Kubernetes Admin)? I'm considering taking the course and getting certified, but a quick search on the internet scared me a little. The course and exam aren't that bad, but the experience of the PSI browser the exam must be taken in apparently is awful: it's a lottery whether it'll work or not, even a passing precheck test does not guarantee anything. People are setting up separate OS installations just for the exam.
Others say that their laptops cannot be used for the exam because of dual-gpu (even on windows).

This sounds like a nightmare.

I'm on LinuxMint 20.3 and I'm actually considering a separate installation of clean ubuntu.

I wonder, has anyone tried taking it? What's the experience? Has anyone tried taking the exam using a Linux (ubuntu?) live-boot?

Why the fuck do we have to use an antivirus that is so intrusive that it has so far wasted my time every day, because it thinks that blocking up random ports is okay. Its a fucking lottery every time I turn on the laptop. I wonder if its going to block port 3306 or 4200 tomorrow...

I have a score difference of two in two posts after eachother on the same page. This is how it must feels to win a lottery without gaining smth. I should buy a ticket. The odds are with me.

So as a personal project for work I decided to start data logging facility variables, it's something that we might need to pickup at some point in the future so decided to take the initiative since I'm the new guy.
I setup some basic current loop sensors are things like gas line pressures for bulk nitrogen and compressed air but decided to go with a more advanced system for logging the temperature and humidity in the labs. These sensors come with 'software' it's a web site you host internally. Cool so I just need to build a simple web server to run these PoE sensors. No big deal right, it's just an IIS service. Months after ordering Server 2019 though SSC I get 4 activation codes 2 MAK and 2 KMS. I won the lottery now i just have to download the server 2019 retail ISO and... Won't take the keys. Back to purchasing, "oh I can download that for you, what key is yours". Um... I dunno you sent me 4 Can I just get the link, "well you have to have a login". Ok what building are you in I'll drive over with a USB key (hoping there on the same campus), "the download keeps stopping, I'll contact the IT service in your building". a week later I get an install ISO and still no one knows that key is mine. Local IT service suggests it's probably a MAK key since I originally got a quote for a retail copy and we don't run a KMS server on the network I'm using for testing. We'll doesn't windows reject all 4 keys then proceed to register with a non-existent KMS server on the network I'm using for testing. Great so now this server that is supposed to connected to a private network for the sensors and use the second NIC for an internet connection has to be connected to the old network that I'm using for testing because that's where the KMS server seems to be. Ok no big deal the old network has internet except the powers that be want to migrate everything to the new more secure network but I still need to be connected to the KMS server because they sent me the wrong key. So I'm up to three network cards and some of my basic sensors are running on yet another network and I want to migrate the management software to this hardware to have all my data logging in one system. I had to label the Ethernet ports so I could hand over the hardware for certification and security scans.

So at this point I have my system running with a couple sensors setup with static IP's because I haven't had time to setup the DNS for the private network the sensors run on. Local IT goes to install McAfee and can't because it isn't compatible with anything after 1809 or later, I get a message back that " we only support up to 1709" I point out that it's server 2019, "Oh yeah, let me ask about that" a bunch of back and forth ensues and finally Local IT get's a version of McAfee that will install, runs security scan again i get a message back. " There are two high risk issues on your server", my blood pressure is getting high as well. The risks there looking at McAfee versions are out of date and windows Defender is disabled (because of McAfee).

There's a low risk issue as well, something relating to the DNS service I didn't fully setup. I tell local IT just disable it for now, then think we'll heck I'll remote in and do it. Nope can't remote into my server, oh they renamed it well that's lot going to stay that way but whatever oh here's the IP they assigned it, nope cant remote in no privileges. Ok so I run up three flights of stairs to local IT before they leave for the day log into my server yup RDP is enabled, odd but whatever let's delete the DNS role for now, nope you don't have admin privileges. Now I'm really getting displeased, I can;t have admin privileges on the network you want me to use to support the service on a system you can't support and I'm supposed to believe you can migrate the life safety systems you want us to move. I'm using my system to prove that the 2FA system works, at this rate I'm going to have 2FA access to a completely worthless broken system in a few years. good thing I rebuilt the whole server in a VM I'm planning to deploy before I get the official one back. I'm skipping a lot of the ridiculous back and forth conversations because the more I think about it the more irritated I get.

Server couldn't take all the demand

I have this fixation with minmaxing core clocks, voltages and their respective curves (P states, basically). I'm still not sure how much it actually improves my experience, but I'll be damned if it isn't fun and interesting to mess around with these numbers and see them climb higher but remain stable.

Currently messing around with a Vega 64. They come in severely overvolted, and I'm trying to get it to retain its performance with less voltage aka less heat. I don't even need to turn on central heating if I'm running stress tests, the bloody thing runs at 80°C! So I guess you can see why undervolting the card is of interest

When you finally find that one working coupon code and feel like you just won the lottery.

Anyone else having so much problems using create-react-app, typescript, styled-components and storybook?
It fells like im playing the lottery here and most of it doesnt reliable work.

Had a customer, for whom we were preparing app for some kind of prize game (you scan a QR code and get a lottery ticket) wanted to name the app "Prize game <name of the client>"...

Executing a docker image without having a minimum amount of trust in it beforehand is like your grandpa clicking on a random lottery winner ad.