spaniard89277

Which LLM do you use? Because I can't see an scraper running daily without being very expensive.

This is completely unrealistic unless you want to burn money.

Not only that, for many people things are going south. I don't really care much about GDP if I have no disposable income nor ability to save money with a normal job. And that's the case for a lot of people in Spain, and I presume other countries too.

Another article on don't go micro services and don't reinvent the wheel.

I wonder if most solutions could be implemented on top of Laravel/Django/Adonisjs without too much effort.

Spain is a lost cause in so many regards...

We can do that. But we can't make new space.

> The purpose of most buildings is to be useful, not to look pretty in pictures.

Architecture is the only art where you're forced to participate. So we could pretty much say the opposite, buildings should, among other things, contribute to make a place nice. Or at least not make it more miserable.

> What do you think Barcelona is?

Most of Spanish cities are built with crap buildings dating above the 60s. Just concrete blocks, in some cities even with just plain grey and humidity stains. It doesn't matter how historic the city is, because the "historic center" it's just a small spot surrounded by uglyness.

In Spain it's not only 60s/70s, pretty much everything build between the 60s and the 90s is not only ugly AF, but very low quality. And Barcelona is not even the worst offender. In northern spanish cities there's a lot of grey in the buildings, with basically no insulation in a humid and cold climate. My GF used to joke it was colder inside her home than outside.

After the new construction code in the early 00s the quality has improved a lot but we still build very ugly buildings.

No wonder why people chose to gather and spend their time where the historial centers are.

Seems architects really refuse to extract some lessons from this.

We dump a lot of subsidies on our factories too. But in the end it doesn't matter, because most people remember that a few years ago you were able to buy chap cars <10k. That mean that modal incomes could afford to have a car.

Now 10k is the price of a second-hand car with lots of potential issues. It got to a point that there are companies that offer pericial services with a guy who goes to review the car with an ODB reader and everything. That wasn't happening here.

But it isn't only the cheap market. Premium brands are pushing a lot of BS and people is taking notice.

You can't squeeze people indefinitely without no repercussion. It's just impossible.

IDK in the rest of Europe, but in Spain there's a growing resentment against euro auto makers as there's a perceived unfair price increase with lots of quality issues and bad customer-facing practices.

Also, everyone is aware they are lobbying the EU for keeping competitors out.

IMO they are building reputational damage and it will last. People is going through hardship and what they see is companies trying to squeeze them and taking advantage.

Every new YT reviewing a car has at least ~5 most upvoted comments complaining about this issues, over and over again, and this has been going for a while.

Same in forums and social media.

I've tried to apply it to parsing HTML as this article into a pretty long pipeline. I'm using DeepInfra with Mistral 8x7B and I'm still unsure if I'm going to use for production.

The problem I'm finding is that the time I wanted to save mantaining selectors and the like is time that I'm spending writing wrapper code and dealing with the mistakes it makes. Some are OK and can deal with them, others are pretty annoying because It's difficult to deal with them in a deterministic manner.

I've also tried with GPT-4 but it's way more expensive, and despite what this guy got, it also makes mistakes.

I don't really care about inference speed, but I do care about price and correctness.

I think JS is fine.

I'm waiting for those job postings of "Junior Engineer with 5 years of experience minimum" in companies with zero pipeline of in-house training.

In most of the west (with few exceptions) is that you should waste at least a decade training yourself with little information of what's useful to learn.

Then, you'll be talked down for being poor and lost. And whatever money you make will be put directly in the hands of your landlords, that profits off scarcity.

It's not a problem in the sense that nowadays we know how to make it. It's a question of money and political build. More housing, more offices, make it pleasent, connect it, etc.

Paris needs a new center. Like many other cities, build more places where people want to flock so you get pressure off the center.

Most new developments are in dead areas because nobody wants to spend time surrounded by ugly, bland and functional architecture.

I've tried to use DuckDB with node in Alpine, and I had problems that needed time to work around.

No problems with SQLite.

I ended up putting the code in Debian because of other problmes I found but the decision was already made.

TLDR is use mistral models on deep infra. Maybe now Mistral updated their prices too.

I use this https://artificialanalysis.ai/

The thing is that this got released to the public after _many_ people said it was ok, internally, at google.

You really have to lost any grip with reality to go for something like that, so I guess they'll rationalize some excuse. That will be interesting to see.

I've had some experiences with selenium and now I'm using puppeteer, and I honestly don't see the problem with JS. It's true that I have not much experience coding but it seems to me that Pupeteer + Flask serving ML to extract data is the cake. Also, being able to play around evaluating expressions in pupeteer, etc, makes it manageable.

Maybe I lack experience but I don't see JS being a barrier.

I would like to know what kind of string work are you doing. I can't imagine being dependent on parsing strings and such, that looks very easy to break, even easier that css selector dance.

Yeah. I managed to abstract a bit the structure but in the end websites change.

It's in my city. Their offices are just a small flat in in a pretty bland neighborhood. I guess most of them work remote.

I'm 36 And also trying to get my foot in. It's very hard, wish I could find someone who gives me an opportunity. I've been interviewed but it's always about the skillset you bring to the table.

Can I throw you against a codebase and be productive quickly? Do you already have experience with this obscure library? I want to build now a website using React + Function Calling + Puppeteer, IDK what else to do to impress people.

Same happening with my gf, who is a graphic designer + UI/UX. She has some experience but only as freelancer, she's never been hired by anyone because it seems never enough.

I teached her HTML + CSS and she's been able to do this by herself (https://sofialenti.com/) so IDK what more a designer can learn. The only thing that's left IMO is to learn JS and be able to code some awesome stuff, but at that point she would be almost an unicorn.

It seems the market is just too saturated, despite the frequent news about we need millions of people who can do X! It seems like a way to get cheap and disposable labor.

>> Richard Morse, a former contributor and one of the support group admins said: “It is so disheartening to see nine years of developing the Duolingo Welsh course being thrown away by disinterested politicians.

I mean, there has to be public employees actually running the thing. It's not like a politician itself wrote anything.

This kind of behavior is so prevalent in big institutions. I wonder if it wouldn't be more effective for the Welsh government to just hand out money to the current mantainers so they at least can take it as a part-time job. The Welsh gov wouldn't be able to take credit for it, but...

Preventing thins would be as easy as to put all courses behind VCS though. Someone fucks up, just roll back.

I like Perplexity but honestly, it lacks "depth" and it doesn't follow instructions very well.

I don't know how it works in the backend but If I ask for a list of sites about some topic what happens it that it gives me the results of a quick search. I've tried to make it really go deep and provide a comprehensive list of links but it never does.

I guess there are multiple problems in such task, like the context of the model, the computational cost of doing such task in the background, etc.

So for me Perplexity is the equivalent of scanning the first page and saving me like 20 minutes/keyword maybe? That's cool enough, but it's not helping when I want to research a topic a bit more deep.

As junior I have coding questions frequently, specially on documentation, and it's better than chat models at it but still pretty limited. And I've spend a fair amount of time trying to make it work, there's a lot of difficult documentation out there and I would love to have a helping hand.

Phind is better at this, the problem with Phind is that is slow, and sometimes you get a blank page. I guess they are overloaded and have less funding than perplexity, thus less money to burn on resources.

What I'm wondering about such companies is how exactly are they going to make a profit. I've tried GPT4 and it's cool and all but still hesitant to pay 20 bucks for it. Maybe if I get a dev job I would think differently but still find myself discarding results often.

I'm a junior, and I have Codeium installed in VSCode. I've found it very distracting most of the times, I don't really understand why so many people uses this kind of assistants.

I find stuff like Phind useful, in the sense that sometimes something happens that I don't understand, and 60% of the times Phind actually helps me to understand the problem. Like finding trivial bugs that I didn't spot because I'm tired, dumb, etc.

On the other hand, with Codeium, I guess it may be useful when you're just churning boilerplate code for some framework, but in my little expericence (writing scrapers and stupid data pipelines & vanilla JS + HTML/CSS) cycling through suggestions is very irritating, specially because many times it doesn't work. Most of the times for stupid reasons, like lacking an argument or something like that, but then it's time you have to spend debugging it.

Another problem I have is that I find there's a common style of JS which consist in daisy-chaining a myriad of methods and anonymous functions, and I really struggle with this. I like to break stuff into lines, name my functions and variables, etc. And so many times code suggestions follow this style. I guess it's what they've been trained on.

Codeium is supposed to learn from this, and sometimes it does, to be fair.

But what I worry the most is that, If I'm a junior and I let this assistants do the code for me ¿How the hell I'm supposed to learn? Because giving Phind context + questions helps me learn or gives me direction to go on find it by myself in the internet, but if the only thing I do is press tab, I don't know how the hell I'm supposed to learn.

I found a couple days ago that many people (including devs) are not using LLMs to get better but it's just a substitute of their effort. Isn't people afraid of this? Not because companies are going to replace you, but it's also a self-reflection issue.

Coding is not the passion of my life, addmitedly, but I like it. I like it because it helps me to make stuff happen and to handle complexity. If you can't understand what's happening you won't be able to make stuff happen and much less to spot when is complexity going to eat you.

Not really. You can use small models for task like text classification etc (traditional nlp) and those run in pretty much anything. We're talking about BERT-like models like distillbert for example.

Now, models that have "reasoning" as an emergent property... I haven't seen anthing under 3B that's capable of making anything useful. The smaller I've seen is litellama and while it's not 100% useless, it's really just an experiment.

Also, everything requires new and/or expensive hardware. For GPU you really are about 1k€ at minumum for something decent for running models. CPU inference is way slower and forget about anythin that has no AVX and preferably AVX2.

I try models on my old thinkpad x260 with 8Gb ram, which is perfectly capable for developing stuff and those small task oriented I've told you about, but even though I've tried everything under the sun, with quantization etc, it's safe to say you can only run decent LLMs with a decent inference speed with expensive hardware now.

Now, if you want task like, language detection, classifying text into categories, etc, very basic Question Answering, then go on HugginFace and try youself, you'll be capable of running most models on modest hardware.

In fact, I have a website (https://github.com/iagovar/cometocoruna/tree/main) where I'm using a small flask server in my data pipeline to extract event information from text blobs I get scraping sites. That runs every day in an old Atom + 4Gb RAM laptop that I use as sever.

Experts in the field say that might change (somewhat) with mamba models, but I can't really say more.

I've been playing with the idea of dumping some money. But I'm 36, unemployed and just got into coding about 1.5 years ago, so until I secure some income I don't want to hit my saving hard, this is not the US where I can land a job easy (Junior looking for job, just in case someone here needs one).

I wonder what's the difference in Airbus.

What adjective would you use to refer as the colony of Moroccans that are next to my neighborhood that sit on public spaces threatening everyone (thus preventing people form using them), being the primary source of crime in the area and displaying their hate towards Spain over and over?

Is it because they were colonized by the french and we had a war many decades ago I have to take a punch in the stomach, or how does this work exactly?