FR version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
71% Positive
Analyzed from 10338 words in the discussion.
Trending Topics
#llms#more#learn#advice#don#learning#llm#build#lot#things
Discussion (238 Comments)Read Original on HackerNews
The reality is that an incredibly small minority of companies in the world do any real training or optimisation. It's unnecessary and inefficient for most purposes unless you are fully dedicated to being an LLM company, and still then it's a struggle. Those few that do train, they spend most of their budget on compute and have relatively small teams.
Getting experience in this field requires having access to very expensive hardware to begin with. And the skills will be quite hard to convert into any real value for someone, leading to a decent income, unless you have a ton of funding from patient investors, or you have decent contacts in Bay Area networks to get hired at the right place.
With all due respect, paulg is in somewhat of a bubble, this is not congruent with the global situation.
For anyone who wants to dork around there is https://github.com/rasbt/LLMs-from-scratch which is something amazing that I think anyone who wants to engineer things around LLMs should at least blast through and read.
It is viable as a toy project, but there are vanishingly few career opportunities.
Not a lot of demand, but also probably not a lot of supply.
Which is paulg's point really
And so I think the idea is more to understand tomorrow ... from first principles.
In the late 80's, as a teenager, I learned x86 assembly and C because that was the only way to squeeze out enough juice from my shitty CGA (and later VGA) card to programm the games/graphics that interested me.
I haven't written assembly in years.
But whatever I did in my career: it helped me and gave me an edge over my peers to have a foundation that is very close to the metal.
I think a lot of people assume that only the big AI labs can do cutting edge research, but there's a strong argument you can do it as part of little tech as well.
Now with LLMs, people write native apps in Rust, and I'd like to think some of them found that there isn't such a huge jump in difficulty they assumed there would be.
Sure you can gradually climb the ladder by demonstrating your skills bit by bit and getting access to more resources. It has very good prospects if you do manage to push through. But it's a hard and risky path, and you will not be able to get any interesting results for the longest time.
For a young middle-class student, it just doesn't make much sense. You can do much more impressive and impactful things with your time without getting into that black hole.
I know how to build an LLM, I know plenty of fellow young engineers that do too. It's really not that complex. But they can't do much with it without capital or access.
Good engineering has never been a bottleneck in this field, it's been all about having access to capital and taking smart but dangerous risks burning it on compute, without much idea of how long you need to keep burning for. There's still no end in sight, some are still managing to convince investors and keep burning, and we are seeing progress, but the business case is still unclear. If you want to get in that game, go ahead, but it's not something I would advice the average young engineer.
I’m 40, and I don’t.I took that abstraction for granted and “left it to the big labs”. However I want to build my own LLM for learning purposes.
On needing big expensive hardware.. necessity is the mother of great innovation. Perhaps 18year olds trying to build their own LLMs in constrained resources environments will result in ground breaking ideas of achieving better intelligence than the one we currently have….
The world needs pragmatic folks who work at a higher abstraction and make LLMs useful, AND also folks who think why not “this other way”? And build newer ways to do fundamental things.
Given the usefulness of current LLMs, I would certainly encourage anybody to try and build their own LLMs, and see what they come up with…
Heck if they build a rack full of old laptops and run something with it that could be done “better” with modern servers, I’d still appreciate the learning running things on those little machines bring.
You can however learn everything you need to know to get on the career ladder as a software engineer on a regular home PC.
The “large” qualifier dates back to pre-transformer language models, where even training a multi-million model was hard due to how poorly it scaled. GPT-2 was a large language model, despite being only 124 millions parameters.
Due to how much high quality data is readily available, anyone can now train a sub-billion (L?)LM on commodity hardware.
And I'm personally convinced that pretty much any enterprise use-case of an LLM (except coding) is better served by a fine-tuned small (<2B) model that is trained specifically on the task, rather than a generalist frontier model, so learning the engineering around fine-tuning is a key skill that companies will realize they need sooner than later.
I just finished fine tuning Gemma e2b for local code completion on my local machine.
This comment just reinforces what the post actually means. We need people that are LLM natives, computing solves itself with time and with scale adjustments
The only jobs that he found he was highly over qualified or paid very little.
In any case, it doesn't look like there's this crazy rush to hire all ML talent, even the one that understand the math and technology deeply.
Maybe people simply don't want math PhDs but something else? Since 1-2 years ago I started doing consulting/freelancing in the ML space, but more on the infrastructure, deployments and similar stuff, as a general purpose developer, and I have a waiting list of clients interested in more work, some of them even trying to recruit me to work for them full-time as well. I'm based in continental Europe, fwiw.
Is it possible to see some of your old works? Personal research?
Citation needed. I'm guessing you're not counting children.
There are plenty of areas were we need people to do this for insurances, banks etc.
AI/ML exists on many levels.
I can see the point. It's unlikely that a 2.4T LLM will be integrated into, say, a pesticide drone. You'll still need some kind of LLMs to achieve maneuvers that "normal" programmings can't achieve.
But what if everything basically turn into that? Essentially, instead of build me a web app to solve X and do Y, build me an LLM to serve X and do Y. (unless the current LLMs are able to do it end-to-end but then they can hardly write coherent software/personal opinion).
1. Both training and optimisation will get significantly cheaper and easier quickly.
2. Politics will probably get even more insane before a potential reprieve on the 20th of Jan 2029.
3. The big AI firms will become part of the surveillance capitalism network, if they're not already.
So I think for self-protection a lot of companies will be looking near to medium term AI independence.
For the time being, unless you truly have millions, the outcome from training will be very net negative, while focusing on building on top of existing AI will yield amazing things if you apply the same talent and effort.
When it does get cheaper, then it will be easier to acquire the skills and experience too, and the struggle you went through by trying to do it now will be somewhat wasted.
Besides, I am well versed in this field, and it is not rocket science. There are plenty of software engineering domains that are a lot more challenging, like high-end graphics, large-scale data engineering or kernel programming. People will learn to train LLMs when people want them to.
In reality, enterprises are happy to offload even risky tasks to others as long as they get some contractual guarantees about their data. Would they like more choice in who to buy from? Yes, but not enough to in-house such a specific discipline.
Fine-tuning a model or LoRA based on the companies data set is more feasible but you're likely going to need several runs as you test/try out different base models, parameters, etc. This is why there are a lot of fine-tuned models on huggingface based on base or instruction-trained models from the larger AI companies that have released open weight models (Microsoft, Google, IBM, Mistral, DeepSeek, Qwen, etc.).
Training is limited on memory first (storing training data and weights) and computation second. Realistically you need to own or rent 2-8 H100/B100 devices or Google's TPUs.
The majority of workflows for a company providing AI capabilities are likely best solved by tailoring a system prompt for the chosen model, evaluating the prompt and model with tools like promptfoo, and then running it on a compute cloud provider (including AWS Bedrock). If the company is big/financially well off enough they could look at buying the hardware needed to run it on their own servers.
For other uses like agentic software development you'd need to spin up a suitable model on a compute cloud provider (or local hardware if the model is small enough) and then tell your IDE/editor to use that model. You would need some way of benchmarking and evaluating the models to see if they are capable of doing the tasks you need. -- There have been some tests done by people on YouTube that suggests that Qwen 3.8 27B is a decent model, but your needs may vary.
You’re missing the point. Understanding how Unity works fundamentally makes you a better Unity dev.
In reality the problem is that it gets blasted out of the water by a much worse architecture trained on 10000x the infrastructure. And while I'm sure the freshly brought in ML student came up with a 10%, even 30% better architecture, it just doesn't matter. (and never mind that even OpenAI hasn't really solved a voice model yet. Try it. It can probably match 2026-quality call centers, but it's no substitute for an actually empowered human)
... and yet, if you look at what hyperscalers are getting paid for ... comfortably more than half the income is training. Which makes no sense on so many levels.
e.g. https://valueaddvc.com/blog/inference-chips-vs-training-chip... (I get it, not great first source, but st
For some very niche cases I think this is probably the case but for the vast majority, the company's data isn't as useful as they think it is or anywhere near the size needed.
I don't know where you are located, but in EU, in China, and yes even in Silicon Valley, the vast majority of companies do not do any real AI engineering. There's nothing wrong with it, it's just not a smart path for most purposes. You can do amazing things without training, and if you try to train, you cannot get anything amazing unless you burn millions.
Very few people can afford to play the long game and cross that dessert. And, sure, you will not get far without good engineering, but good engineering is definitely not sufficient and is not the primary bottleneck.
It would be a good idea for young people to deeply know how these programs work. Not so that they can spend their career building them, but so that they can approach the next class of problems we'll all start trying to solve, with intuition all the way down to the weights and underlying mathematics. And also, to develop a healthy intuition of when "Just LLM it" will not be the right choice.
"Build an OS" wasn't a common university project because we were all expected to go out and work on Windows, but because understanding the bare-metal firmware for a computer helps you deeply understand how to intuit building for a whole class of problems.
I didn't do it because it was useful to me in a practical sense. It's because LLMs are fascinating and I want to know how they work. From that perspective it's been a great experience. I have afirm grasp of the basics. This makes it much easier to understand frontier concepts like compressed latent attention. I can follow the field and understand it.
Not sure I would have got as much out of it at seventeen. I have a lot of background and experience which made it much easier to learn. I wasn't struggling with the linear algebra or with python. I already knew pytorch and neural networks. That helped a lot and I covered these tutorials fast and could skip over large sections. A few evenings and the odd weekend day over a couple of months was enough for me.
For seventeen year olds the tutorials are good enough to make it possible to learn this but it would have taken a lot longer to understand. On the other hand I would have learned a lot more. I think I would have learned a lot of valuable stuff.
However I also think 17 year old me was studying for his A levels and probably this was right choice in terms of maximising future opportunities. I'm not sure I think learning about LLMs instead is sensible. Indeed it might be bad advice. But I can absolutely agree with the sentiment.I think 17 year old me would have wanted to do this too.
Moreover I am not sure it is even good advice? Would you advise a 17 y.o. to learn how transistors work or how to code (i.e. is LLM training the right level in the stack)? LLM training, a discipline where relevant work is already out of reach for 99.999% of budgets really as essential as this post implies?
Personally I don't see the problem, as long as you're aware there is survivorship bias involved here.
What's the alternative really, seek advice from unsuccessful people? That seems worse :)
Personally I do both, read about what worked for people, also read about what didn't work for people, then ignore both and do whatever the fuck I want.
What bucket should I put this advice in?
Agreed, my previous stated "ignore both and do whatever the fuck I want" approach has worked out very well for me in life, people should probably focus on identifying better what their gut tells them, rather than what randoms on the internet thinks and writes.
Intuitively, I would guess that they have a better grasp of what made them fail than successful people have of what made them succeed.
Nonetheless, there are many successful people I would gladly listen to for advice, though they are often successful in a different meaning than what venture capitalists would use (e.g. parents with great kids, managing to keep a healthy work-life balance, happiness, and maybe even having time to spend on some cool hobby project -- you are heros!)
Do both. Get advice from successful and unsuccessful people and take the diff
Seek advice from the averagely successful people, since that is statistically what you're most likely to be.
That’s what you get from listening to “successful people”. You get to learn about all the things they tried that failed, then the things that did work on that 24th try, which was successful.
The “survivorship bias” people always seem to assume that the “survivor” lucked into his fortune on his first try ever, so he can’t have learned anything, so we don’t have to listen to him. But that’s seldom the case.
I’ve written about this before:
https://expatsoftware.com/Articles/survivorship-bias.html
Not sure why would you think so.
Inverse reasoning is very powerful, and unsuccessful people can give you plenty of "don't do this mistake", which the survivors would not even think about.
But how can I know for sure that that particular mistake is actually why they were unsuccessful? Has exactly the same issue as listening only to successful people as they hardly know what actually made them successful most of the time, but they still compose large blog posts with their reasoning for why.
Again, I still think my approach of reading both but then regardless go my own way is the preferable approach, at least for me, ymmv.
https://www.youtube.com/watch?v=3LopI4YeC4I
An advantage that is not "advisable", like being born in january, in a rich country, in an above average family, or just having luck, might have more influence on the outcome than any conscious action. It is almost sure that one-in-a-million level people only edge over the other 999,999 they competed with is just "have more luck".
I think it is. He isn't saying to learn how to train a LLM so that you can go on to train LLMs. He's saying to learn it so that you gain a deep understanding of how LLMs work. Ordinary startups can still benefit from things like training or fine tuning highly specialised smaller models, knowing how to select and configure an appropriate model for the task at hand, knowing what software to use and why, understanding what's going on behind the scenes instead of treating everything like a black box, having a higher level of intuition about LLMs generally, etc.
Most computer science courses do in fact teach things which are lower level than coding, such as how transistors work.
how many of us out here are doing work directly in what we got a degree in? I majored in economics and now I'm a CTO.
I would absolutely advise a 17 yo to learn how to code, understand how transitors work and how to code an llm. even if he never works on llms, you basically end up with a kid with applied knowlege of statistics, math, physics hardware, logic and a whole lot of practice in critical thinking.
Hell yeah. Transistors are pretty awesome.
In 2000 (his era), it would have been really smart to study the source of Linux or Apache. Would have paid dividends over decades. Cuz that knowledge was so rare. The number of people hacking on LLMs now dwarfs the number of people hacking on web servers 30 years ago, by several orders of magnitude.
And if you turn back the clock even more, I mean just even having access to a computer, let alone owning one, would have put you at a massive advantage.
I don't know what to call it. The pioneers should be respected obviously, but at the same time you need to understand that for them, the game wasn't nearly as played out as it is now.
I just don't think you can afford to be dicking around with LLMs like you could afford to dick around with random Linux distros 20 years ago. Too many people willing to do it for free these days.
You don't wanna end up being the 2030 equivalent of a certain SNES emulator developer, or maintainer of a package manager for jailbroken iPhones, I mean the list goes on and on. Being a hacker doesn't automatically give you a path to being rich, or even making a decent living. It hasn't been that way for a while.
I’d probably say something like: do something you enjoy and seems like it might be useful, but accept that the pace of change may mean that whatever you study ends up being irrelevant.
Whatever solution there ends up being to this, it’s not going to be one that an individual 17 year old can implement. We’re past the point where individual good and bad choices matter that much to economic outcomes.
Learning to hack something together in high school using the latest technology (vacuum tubes, radios, microprocessors, web/javascript) has been a common theme in the tech world for generations. With LLMs and online tutorials, this isn't even a difficult suggestion. Do people think learning new tech is somehow wasted effort?
The above is, after all, the whole genesis of the word 'hacker'. We should celebrate that.
I tried to modify the embedding output of bert to make it generate box embeddings instead of point ones. At the time I had access to university provided A100 gpus but even with all that a training run took half a day. Models these days I don't think I can train it in any reasonable time with that much compute.
I can't recall or point out exactly when, but there is a stark before/after moment where the opinions of anything pg went from "Interesting and maybe true in some ways" to what we see today, lots of knee-jerk reactions and hardly any comments about the actual content.
Hazarding a guess, I think the moment Altman became the CEO and later during COVID, the sentiment seemed to have been shifting towards what we see today. But this is all based on hazy memory, rather than looking at the data. I'm sure there is a blog post waiting to be written about analyzing the sentiment of comments to PGs articles on HN, and you'll see a shift somewhere.
Hard disagree. This submission is still being highly upvoted, while another recent post[1] on the harms caused by Graham’s fellows[2], with a fairly tame comment section, has been flagged. That is a constant on HN. It’s not a fluke, it’s as predictable as the sunrise and getting more pronounced.
I’m sure we’re both biased in our perceptions. Mine is that HN in general (certainly more than any other website) used to worship[3] everything he wrote, together with others like Musk, until things started to really go to shit and many eyes have been opened to the effects of the unfettered greed of rich tech guys out of touch with reality.[4]
[1]: https://news.ycombinator.com/item?id=49411762
[2]: A better English word is escaping me.
[3]: That word I choose hyperbolically but deliberately. It definitely was not “interesting and maybe true in some ways”, it was much more hardcore than that.
[4]: That is not “knee-jerk” but a slow realisation still ongoing.
I guess it depends on what submission you look at, previous comment of mine solely based on memory. Now I went to https://news.ycombinator.com/from?site=twitter.com/paulg, clicked "More" a bunch of times, and seems my memory was more or less correct, none of the submissions I clicked on are "pg worship" (hyperbolic or not). Just one example: https://news.ycombinator.com/item?id=19418701
Maybe you need to enable "Show Dead" or something? Pgs articles on HN definitely never was free of any critique in the HN comments, just like any article. Although I do agree with you that it used to be different than it is today, and same with Musk too, and Altman, and probably more individuals, where they were lauded before but now pretty much just mentioning them poisons the conversation.
No. But funnily enough that is a promise by some of the AI cretins and their boosters. Oh yeah best case scenario you learn how to build LLMs for us. We’ll employ you. And then ultimately that just becomes training data for the LLMs to do it themselves.
But why are people cynical? they ask.
Don't many of the commercial ones prevent you from using them to build LLMs?
I would say the reason for the negativity is not because it's a bad idea for a project, or that doing projects in general is a bad idea (it's not!), it's because it's a very specific thing that is not for everyone. The best thing about computing is the low barriers to entry. You can basically work on anything that takes your fancy. So those who are interested in ML will be drawn to learn about LLMs. They don't need anyone to tell them to do it. Telling everyone to do it reminds me of the "just learn to code" stuff of a decade ago. No, please don't, please find something you enjoy.
Being a super rich and an unhappy workaholic, or a super-impressive engineer who wakes up one day at 45 and realizes they regret wasting half their life (I ran into way too many of these) is a much worse fate than "not being rich from your startup" and working a relatively regular job while feeling fulfilled and happy by more than just work.
Especially in the US, which is uniquely bad at this and encourages people to work themselves to death, mental health and work life balance are much more valuable things for 17 year olds to focus on than finding good startup ideas.
In case you think i'm being a bit dramatic, let's look at the state of 17 year old mental health in the heart of Silicon Valley:
"The City of Palo Alto and the Palo Alto Unified School District approved a funded contract to place 24/7 human security guards and monitors at all four local Caltrain grade crossings, including the Churchill Avenue crossing directly adjacent to Palo Alto High School."
(in case it's not obvious, it's because of suicides by high school students)
The 17 year olds do not need advice on better startups, and this situation will never get better if we focus our advice on how to be better at work instead of how to be better at life.
But of course, 10 years ago this wasn't obvious.
Or more reasonably, the same old thing : use Linux, hack a little, why not learn programming basics. But learn to own your technology, fight against centralization of technology. The same old RMS story.
IDK where the tech industry is going, if there will be jobs anymore or not, but what I'm sure (and what have been the case for the last 10-15 years anyway) is that for most tech jobs, having good technical knowledge beyond the basics is pretty useless and will probably not be recognized.
If you can, stay a computer geek if that's your thing, but don't make it your career choice, the Eldorado is behind us.
http://languagemodelbuilder.com teaches you (in a few hours to days) how to build an LLM from scratch. It's entirely free, without accounts, and without data collection.
Thanks a ton for building this.
I’m no paulg, but if you’re reading this - and you’re 17 - just focus on getting into a university and having a good time that you won’t regret later. Play games/sports, make relationships, fall in love, explore.
Then vote for someone who will make the debts go away?
I mean, have you seen the options for people graduating right now? How people are behaving?
Or forget the data, look at how the story of the new future technology is being told. The people making it recognize that it has the potential to put swathes of white collar workers out of jobs, and they are openly talking/warning/PR-ing about it.
People in tech and SV, the places which have a underlying culture of near delusional optimism, are talking about trying to avoid being part of "the permanent underclass".
Gambling is up, and prediction markets are being treated as financial investments. Wall street bets is a thing, and outright speculative investments are the hope people have to get ahead.
This is happening in the USA, forget the weaker or smaller economies.
When people see the future as one massive zero sum game, with no way to win by building, then they are going to change how they plan their future.
Everybody goes to college nowadays and the average white collar has lots of debt and relatively minor financial benefits over a skilled trade worker.
edit: woah, so many people insulted by that. In my bubble and friends, me and another friend are the only people that make very good money compared to non-graduates. Plenty of others opened their shops, went into trades, one learned to tattoo fake eyelashes, one became a (successful) farmer and most make significantly more than the average law/chemist/mathematician/physics/architecture/languages graduates. Sure, the lowest salaries are to be found among the non-graduates too, but I don't see any evidence that graduates make that much more, and that graduating is worth it.
Some answers talking about how "formative college is", but my 25 years old friend with her own shop knows more about real life, business and economy than ivy league MBAs.
Its not good advice for everybody, but it is good advice for a lot of people. What if you want to be a doctor? What if you want to work in R & D? Not everyone enjoys working in a shop or a farm. Also, how old is your friend group? If they are mid twenties you are ignoring the greater scope for advancement in a lot of white collar careers.
> my 25 years old friend with her own shop knows more about real life, business and economy than ivy league MBAs.
Within the narrow limits relevant to her business. How much does she know about macro-economics or financial economics, or scaling up a business? I also suspect you are comparing her to people who went straight on from bachelors to MBA (which is a bad path - study business after having some experience IMO) and lack experience. How will she compare in 10 years time when those people also have real world experience?
Many people are surprised to learn how affordable elite colleges are if you genuinely need financial aid. I had no idea -- was pleasantly surprised when my alma mater took over 80% off of my tuition.
Where I live it's basically free and I still sometimes regret not going into the trades. But I suspect this feeling might mostly be a "grass is greener" thing.
If you’re talking about a place to mature, around others who are at a similar phase of life, also yes.
It is where most people meet their cofounders, for example, even if they don’t found anything until much later.
Plenty of blue collar workers make more than white collar ones, and have a huge debt free head start in life.
A plumber or electrician will make significantly more than the average bank employee or translator or nurse or teacher.
And they will also have an easier time starting a business as many trade workers are self employed and make much more than hired ones.
On an Amiga, I took various public domain text documents from cover disks and counted the probability of the next word given the previous word. Then spat out random sequences of words from it and printed them out. It was called "Splurge". Basically a very very simple single layer statistical language model.
Some of the sentences were randomly not bad sentences, which seemed amazing at the time!
That kind of thing (and Core Wars and Tierra etc) did lead me to getting a job at an artificial life startup at the end of the decade. But that was in turn about 10/15 years too early (no GPUs).
There's some lesson from this about timing, but honestly I've gained the most as a person when I did something that was fun, ethical and gained an audience. A tricky combination.
Putting this in contrast with programming, I learned coding when I was 8, and it was incredibly stimulating to learn because you can quickly iterate and there were thousands of books and YouTube tutorials that dumb everything down and teach you fundamentals. All you needed was a $300 computer, and you can learn nearly anything you want, without being gatekept from this or that because you don't have enough vRAM / an sm_100 GPU.
As a platform engineer being based mainly out of Australia/Hong Kong, opportunities seem to be getting less unless targeting high frequency trading or banking.
It seems like building a startup with the help of some AI tools might be the best bet.
Better to start working with harnesses, evals, statistical analysis, etc. - where you don't need the huge hardware for pre-training etc.
There is very little reason for humans to get all too engrossed in this type of work now, today, with the hope of being good enough at it to command a high salary in 3-5 years. AI can already do it incredibly well, and they can do it persistently and doggedly 24 hours a day.
(Edit: And learn how honest business works)
I think the last one was seeing a skilled electronics repairman do surgery on a CT machine controller.
What I meant was that if you're rich you don't need any trade school or to run a business, you can just sit on your ass living off interest from capital. Only "poor" people need to work for a living.
Might be that these people are from outside the US as well, where things like "honest business" is very much possible today, probably most businesses I interact with AFK on a daily business are "honest businesses".
https://github.com/raiyanyahya/how-to-train-your-gpt
Learn how to make language models from scratch, yes. But learn how to use them, in the context of other machine learning tools, on very small hardware.
When the bubble bursts (and I still tend towards thinking it could burst rather than be deflated in a manageable way), the focus will be on uses of AI that are not like the hyperscalars' products.
People will still be interested in useful AI being added to small things — assistive technologies, home security, garden monitoring, their phones and smartwatches, robotics.
Instead of reductive, reactive make-an-anthropic-competitor advice like this, what about advising 17 year olds to focus on broad, integrated, helpful AI — or on going back through eighty years of history to look at AI projects that failed and reassess them?
learning about llms is not useful so that you can make llms later, you want to learn about it so that you can work on next generation architectures. llms before long i imagine will be left in the dust by ebm / physics oriented models especially that can have an embodied understanding of the world. but a lot of things you learn about them are transferable by doing something like this
The core technoology is pretty basic, developing a rudimentary understanding for why the individual parts work as well as they do is tricky.
2. LLM from 0 to Hero, and nanoGPT by Andrej Karpathy
I first studied them in 2011, and I was like what? Just a bunch of partial derivatives?
I keep looking at AI to check if now it's something else but it keeps being gradient descent.
Okay, it's great that you can perform miracles using gradient descent but that doesn't make it captivating in any way.
I mean first that is already what plenty of 17yo are actually doing, because that is what they do at school or in parascholar activities. There are already countless of such tutorials where you can do that in an afternoon.
The pointless part though is precisely why Amazon and others are hunting for rare books, all the low hanging fruits have been picked already so just training a bigger model will simply mean burning more energy and money. Sure training a small one for the basic principle is a great pedagogical thing, training another one, medium, then maybe a large one, is also good in term of learning the process and architecture, but one should not expect it to be useful out of that context.
Pure players are precisely doing everything they can to corner the market by making their own scale unreachable by others. Smaller players with access to lesser infrastructure are thus betting on different market, e.g. embedded systems.
17yos should definitely build their (L)LMs from scratch and whatever bigger model they can train for free, or for cheap, but they should not expect that to bring them any riches.
You should want to train a LLM from scratch as an intellectual curiosity itch that needs to be scratched.
The idea of learning one hot skill that has a pot of gold waiting at the end of it was a brief moment in time that came and went.
When I was 17, we would have said obviously support vector machines are the future. Neural networks overfit and don't work.
I don’t see why learning how LLMs work is a bad project for a 17 year old.
Optimizing your entire career and the next decade+ of your life on LLMs? Yeah, probably not ideal. It’s almost always a bad idea to make long term decisions based on current trendy things.
And since everyone is using this topic to give their ideal advice to 17 year olds, my advice as a mid-30s guy: seriously consider becoming highly skilled at a specific thing, and don’t be scared off by the idea that it’ll take 5-10-15 years to get there.
When you’re 17-25, the timescale of a decade seems infinite. But it’s really not, and a decade spent “exploring and keeping your options open” sometimes just ends up with you being pretty decent but not amazing at a lot of random things.
Sometimes I wish I had just become a carpenter, chef, electrician, etc. – a specific skill set that leads to mastery over time, rather than the endless exciting-new-thing hamster wheel of working in tech.
I have my opinion on this but I'd like to hear the HN opinion, I will just say one thing:
If you are starting with little knowledge, like a 17 year old would, letting an LLM explain it to you is a terrible idea.
I'm sure they'll get right on powering up their computer from the hamster wheel, Paul.
My heart goes to all the kids out there that didn't get the fair shake let alone fair access to tech that gets these condescending "learn to code/learn to LLM" bootstrappy talks from rich pricks that don't know what life really can be like for a lot of American kids out there.
Learning AI isn't like learning HTML in the 90s then expecting to get a job at a tech company building websites. You can't just "learn how to build LLMs" and expect a frontier lab to hire you so I'd argue this is rather bad advise.
Additionally, unlike web development in the 90s you cant really do anything interesting yourself... All of the interesting/useful stuff will require huge amounts of compute and data so there isn't even much point in learning to start your own thing either.
As someone whose built many of NNs from scratch (hand written code, long before the days of LLMs), it's more or less useless knowledge if I wanted to work in a frontier lab or do anything interesting in the field.
I also think anyone thinking about going into a field which is basically a crossover of CompSci and Maths is absolutely insane right now. Even if you think there is a place for CompSci and Maths post LLMs, there's almost no chance anything you learn today will be relevant to the skills required in say 5-10 years.
If the two current bottlenecks, for this LLM madness that could very well be a bubble, are processing capacity and accuracy (a second processing problem) then what comes next? Isn’t that where young people should be looking or are we just giving up on innovation?
Wait no it’s not, that was always happening.
What’s crazy is that people still believe in it.
Possibilities exist now that did not exist before. Those who do not exploit this are fools.
I do not see any past constructive experience as a waste of time.
And it's basically a weekend project to put transformers together in a ML library and train it.
The follow up comment,train it to play a game also doesn't make sense? Llms Sony really play games and there are better ml approaches to do that?
Even worse when they ask themselves.
Owner of Golf Club Company says I should dedicate my life to golf lmfao.
Sometimes if you want to be heard you must go to the public square, whatever the flags there.
The rest of the post is unintelligible: add some verbs at least.
I'd rather simply write another mnist implementation and check if I really like all that AI stuff at first place. Even then, before going into mature-on-the-way-to-dying tech (LLMs) I'd rather focus on fundamentals - good ols linear models, regressions, stat etc.
If I'd get a buck every time someone said something like this to me when I was in the 13-18 range, I wouldn't have a ton of money, but it's so very annoying when people tell you this.
Regardless if they're "gifted" or not, regardless if you believe in myths like that or not, let children explore what they want to explore, even if you don't understand what it is or why they want to explore that, just let people explore, regardless of age.
It was such a terrible experience being a young kid growing up, with so many adults spending hours trying to convince me to stop sitting in front of the computer so much doing whatever; "why are you even trying to learn that stuff, you have to go to school to understand anything of this" and so much other similar trash.
Sorry, not your fault and I'm borderline trauma-dumping now, but really sad to see this sort of gatekeeping on HN of all places, age is irrelevant to learning ANYTHING, in my humble opinion at least.
Kids, find anything interesting? Jump into it, ignore what adults tell you, and do whatever you feel like, you'll find your place eventually.
This also meant that by the time I was actually offered to take a programming class in school (junior year of HS), I had already been able to self-teach myself well beyond what that class was covering, thanks to just working on random projects that scratched an itch I had at the time, looking up anything I didn't know or understand, and internalizing those concepts over time.
In short though, I definitely agree, young kids and teens (and also, frankly, adults too!) should be encouraged to explore things that they have a passion for, without being told 'you need to go to school for this' or 'you cant understand this at your age'
There still will be varyy small number of outliers among youngsters who'd be able to extract tremensous value from such an excercise, but for most that'd be _IMO_ waste of of time, with illusion of understanding w/o actually having any.
Same! I just happened to disagree with your opinion, and frankly, I'd say trying to gatekeep what people learn is closer to "borderline irresponsible" compared to asking people to build/learn/do X.
> youngsters who'd be able to extract tremensous value from such an excercise
But they're youngsters, who are about "extracting value"? Life is about fun, not extraction, not value, not avoiding waste of time but literally enjoy what you do, nothing is more important (IMO).
Then who knows, doing fun stuff sometimes lead to useful stuff, like in my life. But if you only think about "extracting most value for time spent" or similar "optimization strategies", then you'd never discover this part of life.
Because I can?. JK. Because that was my experience, of someone who is 2.5 older than 17?
> For some (many?) people, a 'proper' understanding develops _after_ the exploration.
I am afraid you have a too confrontational attitude here, but I'll answer anyway: because I do not believe you can simply "explore" such complex topics like building an LLMs. You'd simply be unable to build LLM drom scratch, unless you'd call cargo-cult chaining magic numpy incantations you've taken from Karpathy's tutorials "exploring".
If I were in "exploratory" state of mins, I'd rather go from entirely different side - I'd try playing with LoRA-ing existing small LLMs, such as venerable 2 y.o. Mistral Nemo, to get "feeling" for what training is and how hyperameters influence the process.
That's what I would call exploring.
I too started exploring programming as a teen by cargo culting. Fooling around and getting results is what made it fun. Understanding came later.
I attempted many projects at a young age that I was absolutely not equipped for. The result of the attempts more often than not left me equipped, every time it left me better off. This is terrible advice.
Transformers are difficult to understand even to people with strong ML background, let alone a teenager.
That 17yo would have already built many uncommon bases, and would build further.
That is, a 17yo with proper mentality.
Can you provide an example?
Either way, this isn't really advice for 17 year olds. Pg is thinking out loud about the pathways for founders.
>Whoa. I’m 19 and I trained a 100M language model from scratch. Did a v2 now with a new SFT experiment to see if I can get better results on same size.