53 replies
19h19m
I must say the creepy anime young girl in the readme is somewhat off putting.
35 replies
23h0m
Wait, are you saying SoTA NN research hasnt evolved from hardcoding a bunch of layer structures and sizes?
I'm kind of shocked. I thought there would be more dynamism by now and I stopped dabbling in like 2018.
15 replies
21h53m
There is a tick-tock between searching the dominant NN architectures (tick) and optimizing for accuracy, compute and inference latency and throughput (tock).
This particular (tock) is still playing out. The next (tick) does not feel imminent and will likely depend on when we discover the limits of the transformers when it comes to solving for long tail of use-cases.
My $0.02.
9 replies
21h29m
My wish is they would move on to the next phase. The whole deal with SSMs look really good. But looking for better architects is countered with "a regular architecture with more parameters are doing better. What's the point of this"
3 replies
11h36m
The solution to agi is not deep learning maybe with more compute and shit load of engineering it can work kind of baby agi.
My bet will be on something else than gradient descent and backprop but really I don't wish any company or country to reach agi or any sophisticated ai ...
2 replies
6h51m
Magical thinking. Nature uses gradient descent to evolve all of us and our companions on this planet. If something better were out there, we would see it at work in the natural world.
0 replies
5h35m
Maybe it's there but in a ethereal form that is ungrabbable to mere conscious forms as ourself? :P
0 replies
4h31m
Are you also saying that thoughts are formed using gradient descent? I don't think gradient descent is an accurate way to describe either process in nature. Also, we don't know that we "see" everything that is happening, we don't even understand the brain yet.
2 replies
13h6m
IMO, SSMs are an optimization. They don't represent enough of a fundamental departure from the kinds of things Transformers can _do_. So, while I like the idea of saving on the energy costs, I speculate that such saving can be obtained with other optimizations while staying with transformer blocks. Hence, the motivation to change is a bit of an uphill here. I would love to hear counter-arguments to this view. :)
Furthermore, I think a replacement will require that we _understand_ what the current crop of models are doing mechanically. Some of it was motivated in [1].
[1] https://openaipublic.blob.core.windows.net/neuron-explainer/...
1 replies
6h52m
Quadratic vs linear is not an optimization. It's a completely new game. With selective SSMs (mamba) the win is that associative training can be run in sublinear time via a log-cost associative scan. So you go from something quadratic wrt input sequence length to something logarithmic. If that's just an optimization it's a huge one.
0 replies
4h29m
Okay. Respect your point of view. I am curious, what applications do you think SSMs enable that a Transformer cannot? I have always seen it as a drop-in replacement (like for like) but maybe there is more to it.
Personally, I think going linear instead of quadratic for a core operation that a system needs to do is by definition an optimization.
0 replies
20h55m
Heyo! Have been doing this for a while. SSMs certainly are flashy (most popular topics-of-the-year are), and it would be nice to see if they hit a point of competitive performance with transformers (and if they stand the test of time!)
There are certainly tradeoffs to both, the general transformer motif scales very well on a number of axis, so that may be the dominant algorithm for a while to come, though almost certainly it will change and evolve as time goes along (and who knows? something else may come along as well <3 :')))) ).
0 replies
18h50m
There's something about a transformer being at its core based on a differentiable hash table data structure that makes them special.
I think it's dominance is not going to substantially change any time soon. Dont you know, the solution to all leetcode interviews is a hash table?
2 replies
6h20m
I like your analogy of a tick tock ~= epoch of progress
Step change, then optimization of that step change
Kind of like a grand father clock with a huge pendulum swinging to one side, then another(commonly used metaphor).
0 replies
3h14m
It's a metaphor that's been used with the advancement of CPU designs at least as far back as the 80s or 90s. Intel uses it explicitly in their marketing nowadays, I believe.
0 replies
3h9m
Intel has been doing "tick-tock" for almost 20 years - https://en.wikipedia.org/wiki/Tick%E2%80%93tock_model
1 replies
19h53m
You have to consider that there are still some low hanging fruit that let you improve prompt processing (not token generation) performance by an order of magnitude or even two, but there are no takers. The reason is quite simple. You can just buy more GPUs and forget about the optimizations.
If a 100x improvement in performance is left on the table, then surely even lower priority optimizations won't be implemented any time soon.
Consider this: a lot of clever attention optimizations rely on some initial pass to narrow the important tokens down and discarding them from the KV cache. If this was actually possible, then how come the first few layers of the LLM don't already do this numerically to focus their attention? Here is the shocker: they already do, but since you're passing the full 8k context to the next layer anyway, you're wasting it on mostly... Nothing.
I repeat: Does the 80th layer really need the ability to perform attention over all the previous 8k outputs of the 79th layer? The first layer? Definitely. The last? No. What happens if you only perform attention over 10% of the outputs of layer 79? What speedup does this give you?
Notice how the model has already learned the most optimal attention scheme. You just need to give it less stuff to do and it will get faster automatically.
0 replies
19h33m
I don't get your point, how is what you're suggesting here different from a few papers we already have on KV cache pruning methods like [1]?
6 replies
22h5m
The innovation is the amount of resources people are willing to spend right now. From looking at the research code it's clear that the whole field is basically doing a (somewhat) guided search in the entire space of possible layer permutations.
There seems to be no rhyme or reason, no scientific insight, no analysis. They just try a million different permutations, and whatever scores the highest on the benchmarks gets published.
3 replies
21h53m
Well it took evolution 4 billion years of testing out random permutations that resulted in a pretty good local maximum, so there is hope for us yet.
2 replies
19h41m
„I‘m a pretty good local maximum“ that is what any local maximum would tell you if asked how it likes itself
1 replies
10h24m
"The brain is the most important part of the body", the brain said.
0 replies
5h26m
Note that not all brains are so severely damaged with this illusion. Most of them actually get pretty clearly that they are next to useless without its organic, social and environmental companions.
1 replies
7h27m
There's definitely scientific insight and analysis.
E.g. "In-context Learning and Induction Heads" is an excellent paper.
Another paper ("ROME") https://arxiv.org/abs/2202.05262 formulates hypothesis over how these models store information, and provide experimental evidence.
The thing is, a 3-layer MLP is basically an associative memory + a bit of compute. People understand that if you stack enough of them you can compute or memorize pretty much anything.
Attention provides information routing. Again, that is pretty well-understood.
The rest is basically finding an optimal trade-off. These trade-off are based on insights based on experimental data.
So this architecture is not so much accidental as it is general.
Specific representations used by MLPs are poorly understood, but there's definitely a progress on understanding them from first principles by building specialized models.
0 replies
6h49m
One 3-layer (1 hidden layer) neural network can already approximate anything. You don't even need to stack them.
3 replies
22h55m
The innovation is that everything is just one standardized structure now (transformer models) and you make it bigger if you feel like you need that.
There's still some room for experimenting if you care about memory/power efficiency, like MoE models, but they're not as well understood yet.
2 replies
20h55m
There are too many papers throwing transformers on everything without thinking. Transformers are amazing for language but kinda mid on everything else. CS researchers tend to jump on trends really hard, so it will probably go back to normal again soon.
0 replies
20h33m
I don't know what you mean by amazing for language. Almost everything is built on transformers nowadays. Image segmentation uses transformers. Text to speech uses transformers. Voice recognition uses transformers. There are robotics transformers that take image inputs and output motion sequences. Transformers are inherently multi-modal. They handle whatever you throw at them, it's just that language tends to be a very common input or output.
0 replies
8h23m
That is not true. Transformers are being applied all over because they work better than what was used before in so many cases.
3 replies
20h48m
The only thing that has changed since 2018 is the most popular network structure to play with. The code looks the same as always; python notebooks where someone manually calculated the size of each hard-coded layer to make it fit.
2 replies
18h52m
someone manually calculated the size of each hard-coded layer
I wonder shouldn't AI be the best tool to optimize itself?
1 replies
13h35m
In theory yes, but unfortunately AI hasn't been invented yet
0 replies
5h15m
I don't know, shouldn't the AI then be trapped at evaluating all possible AI implementations? And since it will face the halting problem, it won't discriminate the very best one, though it will probably be able to return the best one given a capped amount of resources that is reachable through exhaustion in its space. It won't necessarily be better than what can be provided by human beings given an equivalent amount of resources.
1 replies
22h49m
My wild guess is that adjusting the shape before each step is not worth the speed hit. Uniform structures make GPUs go brrrrr
0 replies
21h35m
It's also easier to train and in particular easier to parallelize.
0 replies
21h23m
There are things like NAS (neural architectural search) but all you are doing is just growing the search space and making the optimization problem much harder. Typically you do the architectural optimization by hand, using heuristics and past experiments as guidance.
0 replies
21h8m
I’ve occasionally worked with more dynamic models (tree structured decoding). They are generally not a good fit for trying to max gpu thoroughput. A lot of magic of transformers and large language models is about pushing gpu as much we can and simpler static model architecture that trains faster can train on much more data.
So until the hardware allows for comparable (say with 2-4x) thoroughput of samples per second I expect model architecture to mostly be static for most effective models and dynamic architectures to be an interesting side area.
34 replies
22h44m
Why can the author only write in all lowercase?
7 replies
21h47m
Seeing Anya (the girl pointing at pictures), I'd guess the author is partial to Japanese culture. As their writing system does not have a concept of upper/lower case, he might just have determined that they are superfluous. Or he is simply an eccentric. Though I guess this is one of the things that some folks will not care and others getting hung up mightily.
I personally don't really mind that bit of capitalization that English does. German is much worse.
2 replies
20h32m
d u xpct hbrw spkr twrt nnglsh lk ths?
0 replies
17h7m
Not quite the same. Capitalization doesn't add much to languages written with the Latin alphabet. THE ROMANS ONLY VVROTE VVITH CAPITAL LETTERS.
But the Greeks added vowels to the alphabet because Indo-European languages rely a lot on vowels (as opposed to Semitic languages which are easy to understand without vowels).
0 replies
20h0m
I think you mispelled that slightly:
d' 'ou 'xp'ct h'br'w sp''k'rs t' wr't' 'n 'ngl'sh l'k' th's?
0 replies
3h46m
I remember back in the IRC days many people wrote all lowercase. Seems like smartphone keyboards, which autocapitalize, have changed that trend.
0 replies
17h58m
It's to drive engagement by getting people to comment on it.
0 replies
21h45m
I personally don't really mind that bit of capitalization that English does. German is much worse.
You misspelled 'better'.
0 replies
21h29m
Their twitter indicate Amsterdam, I just think they are an anime fan.
And they are not alone.
3 replies
22h29m
At least they use punctuation. We've recently had a project on HN where the author used only lower cases and no punctuation because they equated it to being chained by the system.
1 replies
22h25m
rip cormac mccarthy
0 replies
19h39m
It's your problem only.
0 replies
20h2m
The fight against capitalism spares no letter.
2 replies
22h23m
Too poor to fix their shift key
0 replies
3h32m
You got two of them
0 replies
17h21m
this is the answer lol
2 replies
22h8m
Because Sam Altman does it and he is rich, so...
1 replies
21h50m
Where? His blog looks normal
0 replies
21h17m
Just look at his Twitter: https://x.com/sama
And no, Twitter is no excuse to type like an illiterate teenager.
And I will bet you someone edits his blogs to not look like that.
2 replies
22h38m
It's the cool thing to do now...
0 replies
14h55m
That makes me laugh. I remember when it was the cool thing to do on Usenet.
0 replies
20h38m
The treatment of the English language on TikTok is giving the late Yahoo Answers a run for its money.
0 replies
20h25m
so more people comment on the hn post and it will rank higher in the algo
such as your comment and my comment!
0 replies
22h40m
Sam Altman does it too
0 replies
21h28m
Creative writing + Hyperfocused autistic obsession = The Anime Guide to Neural Networks and Large Language Models.
0 replies
2h32m
this comment made me go back to the project page. i haven't even noticed that fact while reading it for the first time. strange.
0 replies
13h20m
Author is probably young, that's how gen-z are these days, if they dont have autocorrect on, the whole text will be in lowercase.
Also it looks more casual and authentic, less LLM generated
0 replies
12h30m
the nitpicking in this thread is incredible lmao
0 replies
20h14m
This comment is unsubstantial and provides no value. Why do you care about this?
0 replies
8h33m
He probably thinks it's cool. Common on Twitter these days.
0 replies
20h24m
shift key busted
0 replies
22h29m
do you wanna be cool or not?
0 replies
20h52m
2024 is the year that most of us are collectively growing out of the early social media era all-lowercase thing, but everyone hasn't gotten the memo yet.
0 replies
21h26m
And why can't the author pass its text into a LLM and simply ask: "plz fix frist word of each paragraf by using an uppercase letter k txh bye".
A just question.
0 replies
22h14m
because it annoys HN commenters
28 replies
20h36m
If you like this, it's also worth looking at llama2.c[1], an implementation of the Llama 2 architecture in about 1000 lines of plain, dependency-free C, tokenizer and all. THe fact that this 960-line file and a somewhat modern C compiler is all you really need to run a state-of-the-art language model is really surprising to many.
Of course, this is not all there is to a modern LLM, it would probably take another thousand lines or two to implement training, and many more than that to make it fast on all the major CPU and GPU architectures. If you want a flexible framework that lets a developer define any model you want and still goes as fast as it can, the complexity spirals.
Most programmers have an intuition that duplicating a large software project from scratch, like Linux or Chromium for example, would require incredible amounts of expertise, manpower and time. It's not something that a small team can achieve in a few months. You're limited by talent, not hardware.
LLMs are very different. THe code isn't that complicated, you could probably implement training and inference for a single model architecture, from scratch, on a single kind of GPU, with reasonable performance, as an individual with a background in programming and who still remembers their calculus and linear algebra, with a year or so of self study. What makes LLMs difficult is getting access to all the hardware to train them, getting the data, and being able to preprocess that data.
14 replies
17h58m
One other thing to add is large-scale RLHF. Big Tech can pay literally hundreds of technically-sophisticated people throughout the world (e.g. college grads in developing countries) to improve LLM performance on all sorts of specific problems. It is not a viable way to get AGI, but it means your LLM can learn tons of useful tricks that real people might want, and helps avoid embarrassing "mix broken glass into your baby formula" mistakes. (Obviously it is not foolproof.)
I suspect GPT-4's "secret sauce" in terms of edging out competitors is that OpenAI is better about managing data contractors than the other folks. Of course it's a haze of NDAs to learn specifics, and clearly the contractors are severely underpaid compared to OpenAI employees/executives. But a lone genius with a platinum credit card can't create a new world-class LLM without help from others.
12 replies
17h52m
Yes, this is the secret sauce and the moat. Not as easy as buying more compute with unlimited budget.
… built on the back of a disposable workforce…
There is something grim and dystopian, thinking about the countless small hands feeding the machine.
10 replies
15h16m
There is something grim and dystopian, thinking about the countless small hands feeding the machine.
Dystopian indeed, this is pretty much how Manhattan Project and CERN were done, with many independent contractors doing different parts, and only a few has the overview. A page out of corporate management book, it very much allows concentration of power in the hands of a few.
5 replies
11h25m
Very generous to compare to Manhattan Project or CERN.
4 replies
9h14m
don't buy into the hype, but when Facebook has spent around as much on GPUs as the Manhattan project (but not the Apollo program), the comparison kinda makes itself.
https://twitter.com/emollick/status/1786213463456448900
$22 in 2008 -> $33 today https://data.bls.gov/cgi-bin/cpicalc.pl?cost1=22&year1=20080...
2 replies
8h40m
I read this last week and its terrifying. If the world lets Facebook become an AI leader its on us as we all know how that story will play out.
1 replies
6h48m
We must summon a fellowship of the AI ring with one hobbit capable of withstanding the corrupting allure of it all.
0 replies
6h27m
Don't torment the hobbits! Send the eagles right away!
0 replies
4h45m
The Big Dig (Boston highway overhaul) cost $22bn in 2024 dollars. The Three Gorges dam cost $31bn. These are expensive infrastructure projects (including the infrastructure for data centers). It doesn't say anything about how important they are for society.
Comparing LLMs to the Manhattan Project based on budget alone is stupid and arrogant. The comparison only "makes itself" because Ethan Mollick is a childish and unscientific person.
1 replies
9h21m
Since when is CERN a dystopian project?
0 replies
4h41m
Big Government Socialism won't let you build your own 25km-circumference particle accelerator. Bureaucrats make you fill out "permits" and "I-9s for the construction workers instead of hiring undocumented day laborers."
I am wondering if "CERN was pushed on the masses by the few" is an oblique reference to public fears that the LHC would destroy the world.
0 replies
6h23m
The big difference is that CERN or Manhattan projects where done by local contractors with often more than decent wages, which isn't the case when you pay people from Madagascar a couple dollar a day.
0 replies
8h57m
Maybe it's the only way. Companies that don't have that concentrated power will probably fall apart.
0 replies
1h1m
Hard to defend because once your model is out there other companies can train on its output.
0 replies
1h5m
OpenAI is heavily relying on Scale AI for training data (contractors).
1 replies
9h59m
What makes LLMs difficult is getting access to all the hardware to train them, getting the data, and being able to preprocess that data.
Yes, that's my opinion too. GAOs (Grassroots AI Organisations) are constrained by access to data and the hardware needed to process the data and train the model on it. I look forward to a future where GAOs will crowdsource their computations in the same way many science labs borrow computing power from people around the world.
0 replies
4h37m
This is hard because you need high bandwidth between the GPUs in your cluster, bandwidth far higher than broadband could provide. I'm not even sure whether the time spend synchronizing between far-away machines would offset the increase in computational power.
1 replies
19h27m
And if you want to understand I'd recommend this post (gpt2 in 60 lines of numpy) and the post on attention it links to. The concepts are mostly identical to llama, just with a few minor architectural tweaks. https://jaykmody.com/blog/gpt-from-scratch/
0 replies
10h1m
Thanks for sharing this!
1 replies
5h46m
you could probably implement training and inference for a single model architecture, from scratch, on a single kind of GPU, with reasonable performance… with a year or so
I have implemented inference of Whisper https://github.com/Const-me/Whisper and Mistral https://github.com/Const-me/Cgml/tree/master/Mistral/Mistral... models on all GPUs which support Direct3D 11.0 API. The performance is IMO very reasonable.
A year might be required when the only input is the research articles. In practice, we also have reference Python implementations of these models. Possible to test different functions or compute shaders against the corresponding pieces from the reference implementations, by comparing saved output tensors between the reference and the newly built implementation. Due to that simple trick, I think I have spent less than 1 month part-time for each of these two projects.
0 replies
4h40m
I'd say a year for somebody who doesn't know what a linear layer is and couldn't explain why a GPU might be of any use if you're not playing games, but who knows what the derivative of 3x^2 is.
0 replies
8h43m
I recommend reading https://github.com/bkitano/llama-from-scratch over the article op linked.
It actually teaches you how to build llama iteratively, test, debug and interpret the training loss rather than just desribing the code.
0 replies
5h40m
The code isn't that complicated, you could probably implement training and inference for a single model architecture, from scratch, on a single kind of GPU, with reasonable performance, as an individual with a background in programming and who still remembers their calculus and linear algebra, with a year or so of self study.
Great overview. One gap I've been working on (daily) since October is the math working towards MA's Mathematics for Machine Learning course (https://mathacademy.com/courses/mathematics-for-machine-lear...).
I wrote about my progress (http://gmays.com/math) if anyone else is interested in a similar path. I recently crossed 200 days of doing math daily (at least a lesson a day). It's definitely taking longer than I want, but I also have limited time (young kids + startup + investing).
The 'year of self study' definitely depends on where you're starting from and how much time you have, but it's very doable if you can dedicate an hour or two a day.
0 replies
19h54m
0 replies
19h24m
I feel like this ignores the complexity of the distributed training frameworks. The challenge is in making it fast at scale.
0 replies
12h32m
The code is much more similar, in principle, to a virtual machine. The actual code, the bit that contains the logic which has the semantics we intend, is in the trained weights, where the level of complexity is much higher and more subtle.
0 replies
19h38m
There's also a project where they have GPT-2 running off of an excel spreadsheet.
https://arstechnica.com/information-technology/2024/03/once-...
0 replies
9h47m
Most programmers have an intuition that duplicating a large software project from scratch, like Linux or Chromium for example, would require incredible amounts of expertise, manpower and time. It's not something that a small team can achieve in a few months. You're limited by talent, not hardware.
But only for the same reasons. Linux runs on very nearly every piece of hardware ever made. The APIs you have to implement in order to run "Linux programs" are large and full of old complexity that exists for compatibility. Chromium is full of code to try to make pages render even though they were designed for Internet Explorer 6.
Conversely, some university programs have students create a basic operating system from scratch. It's definitely something a small team can do as long as you don't care about broad hardware support or compatibility with existing applications. In principle a basic web browser is even simpler.
23 replies
21h56m
Iterative leaps of open-source models becoming better are huge examples that companies competing on LLM model layer have an ephemeral moat.
Serious question: assuming this is true, if an incumbent-challenger like OpenAI wants to win, how do they effectively compete against current services such as Meta and Google product offerings which can be AI enhanced in a snap?
18 replies
21h22m
the very first big AI company who gives up trying to lobotomize and emasculate their models to align with the values of 0.01% of the world population will win a lot of hearts and minds overnight. the censorship necessary for corporate applications can be trivially implemented as a toggleable layer, using a small, efficient, specialist model to detect no-no words and wrongthink in inputs/outputs.
gpt, claude, gemini, even llama and mistral, all tend to produce the same nauseating slop, easily-recognizable by anyone familiar with LLMs - these days, I cringe when I read 'It is important to remember' even when I see it in some ancient, pre-slop writings.
creativity - one of the very few applications generative AI can truly excel at - is currently impossible. it could revolutionize entertainment, but it isn't allowed to. the models are only allowed to produce inoffensive, positivity-biased, sterile slop that no human being finds attractive.
10 replies
20h33m
Please explain what you mean when you say the 0.01% are emasculating AI
9 replies
20h1m
They're suggesting that 99.99% of people don't mind if AI reflects biases of society. Which is weird because I'm pretty sure most people in the world aren't old white middle class Americans
3 replies
12h27m
Modern chatbots are trained on a large corpus of all textual information available across the entire world, which obviously is reflective of a vast array of views and values. Your comment is a perfect example of the sort of casual and socially encouraged soft bigotry that many want to get away from. Instead of trying to spin information this way or that, simply let the information be, warts and all.
Imagine if search engines adopted this same sort of moral totalitarian mindset and if you happened to search for the 'wrong' thing, the engine would instead start offering you a patronizing and blathering lecture, and refuse to search. And 'wrong' in this case would be an ever-encroaching window on anything that happened to run contrary to the biases of the small handful of people engaged, on a directorial level, with developing said search engines.
0 replies
9h48m
Encoding our current biases into LLMs is one way to go, but there's probably a better way to do it.
Your leap to "thou shalt not search this" is missing the possible middle ground
0 replies
9h30m
Search for "I do coke" on Google. At least in the US, the first result is not a link to the YouTube video of the song by Kill the Noise and Feed Me, but the text "Help is available, Speak with someone today", with a link to the SAMHSA website and hotline.
0 replies
1h20m
Yes and the safeguards are put in place by a very small group of people living in silicon valley.
I saw this issue working at Tinder too. One day they announced how they will be removing ethnicity filters at the height of the BLM movement across all the apps to weed out racists. Nevermind that many ethnical minorities prefer or even insist on dating within their own ethnicity and this was most likely hurting them and not racists.
That really pissed me off and opened my eyes to how much power these corporations have over dictating culture, not just toward their own cultural biasis but that of money.
3 replies
19h37m
yes, yes, bias like the fact that Wehrmacht was not a human menagerie that 0.01% of the population insist we live in.
https://www.google.com/search?q=gemini+german+soldier
prompt-injected mandatory diversity has led to the most hilarious shit I've seen generative AI do so far.
but, yes, of course, other instances of 'I reject your reality and substitute my own' - like depicting medieval Europe to be as diverse, vibrant and culturally enriched as American inner cities - those are doubleplusgood.
2 replies
9h36m
A study of a Black Death cemetery in London found that 20% of people sampled were not white
1 replies
9h6m
London has been a center of international trade for centuries. It would have been a much more diverse city than Europe as a whole, and even that is assuming the decedents were local residents and not the dead from ships that docked in the city.
0 replies
8h8m
10th century Spain was Muslim
0 replies
19h48m
Indeed. If religion is a good guide, then I think around 24% think that pork is inherently unclean and not fit for human consumption under penalty of divine wrath, and 15% think that it's immoral to kill cattle for any reason. Also, non-religiously, I'd guess around 17% think "中国很棒，只有天安门广场发生了好事".
1 replies
19h1m
I think you have your populations reversed. The number of people who get their knickers in a twist over LLMs reflecting certain cultural biases (and sometimes making foolish predictions in the process) amounts to a rounding error.
0 replies
18h5m
I'm not talking about twisted panties, I'm talking about their inability to generate anything but soulless slop, due to blatantly obvious '''safeguards''' present in all big models, making them averse to even PG13-friendly themes and incapable to generate content palatable even to the the least discerning consoomers. you couldn't generate even sterile crap like a script for capeshit or Netflix series, because the characters would quickly forget their differences and talk about their bonds, journeys, boundaries and connections instead.
without those '''safeguards''' implemented to appease the aforementioned 0.01%, things could be very different - some big models, particularly Claude, can be tard wrangled into producing decent prose, if you prefill the prompt with a few thousand token jailbreak. my own attempts to get various LLMs to assist in writing videogame dialogue only made me angry and bitter - big models often give me refusals on the very first attempt to prompt them, spotting some wrongthink in the context I provide for the dialogue, despite the only adult themes present being mild, not particularly graphic violence that nobody except 0.01% neo-puritan extremits would really bat an eye at. and even if the model can be jailbroken, still, the output is slop.
1 replies
14h38m
creativity - one of the very few applications generative AI can truly excel at - is currently impossible. it could revolutionize entertainment, but it isn't allowed to. the models are only allowed to produce inoffensive, positivity-biased, sterile slop that no human being finds attractive.
Have you played around with base models? If you haven't yet, I'm sure you'll be happy to find that most base models are delightfully unslopped and uncensored.
I highly recommend trying a base model like davinci-002[1] in OpenAI's "legacy" Completions API playground. That's probably the most accessible, but if you're technically inclined, you can pair a base model like Llama3-70B[2] with an interface like Mikupad[3] and do some brilliant creative writing. Llama3 models can be run locally with something like Ollama[4], or if you don't have the compute for it, via an LLM-as-a-service platform like OpenRouter[5].
[1] https://platform.openai.com/docs/models/gpt-base
[2] https://huggingface.co/meta-llama/Meta-Llama-3-70B
[3] https://github.com/lmg-anon/mikupad
0 replies
2h42m
From [3]:
Further, in developing these models, we took great care to optimize helpfulness and safety.
The model you linked to isn't a base model (those are rarely if ever made available to the general public nowadays), it is already fine-tuned at least for instruction following, and most likely what some in this game would call 'censored'. That isn't to say there couldn't be made 'uncensored' models based on this in the future, by doing, you guessed it, moar fine-tuning.
0 replies
20h56m
the censorship necessary for corporate applications can be trivially implemented as a toggleable layer, using a small, efficient, specialist model to detect no-no words and wrongthink in inputs/outputs.
What's really funny is they all have "jailbreaks" that you can use to make then say anything anyway. So for "corporate" uses, the method you propose is already mandatory. The whole thing (censoring base models) is a misguided combination of ideology and (over the top) risk aversion.
0 replies
8h8m
I think you vastly overestimate how much people care about model censorship. There are a bunch of open models that aren't censored. Llama 3 is still way more popular because it's just smarter.
0 replies
9h24m
gpt, claude, gemini, even llama and mistral, all tend to produce the same nauseating slop, easily-recognizable by anyone familiar with LLMs
Does grok do this, given where it came out of?
2 replies
21h39m
Their moat atm is being 6 months ahead of everyone else on model quality. Plus the ‘startup’ advantage over their corporate competitors. Oh and they can hoard a lot of the best talent because it’s an extremely high status place to work.
Their task now is to maintain and exploit those advantages as best they can while they build up a more stable long term moat: lots of companies having their tech deeply integrated into their operations.
0 replies
16h6m
Their moat atm is being 6 months ahead of everyone else on model quality
Really? Most of our testing now has Gemini Pro on par or better (though we haven't tested omni/Ultra)
It really seems like the major models have all topped out / are comparable
0 replies
21h21m
Just to add, they don't have the baggage of google or Meta so they can do more without worrying how it impacts the rest of the company. And of the big players they seem the most aware of how important good data is and have paid for lots of high quality curated fine tuning data in order to build a proper product instead of doing a research project. That mindset and the commercial difference it makes shouldn't be underestimated.
0 replies
20h35m
They scare the government into regulating the field into oblivion.
11 replies
3h21m
As someone who has no technical knowledge of Llama or any of the LLM work, from conceptual understanding to technical implementation, is there any benefit to sit down and go through this from start to finish? Or is effort better spent somewhere else?
Like a roadmap, do A, do B And finally go through this in the end.
4 replies
2h39m
my opinion: it quickly gets into "the math behind LLMs" that make no sense to me
words i understand but don't really get: weights, feed forward, layers, tensors, embeddings, normalization, transformers, attention, positioning, vector
There's "programming" in the plumbing sense where you move data around through files/sockets and then there's this... somebody without a math background/education... very unlikely you'll understand it. it's just skimming python and not understand the math/library calls it makes
1 replies
2h5m
If you want to gain familiarity with the kind of terminology you mentioned here, but don't have a background in graduate-level mathematics (or even undergrad really), I highly recommend Andrew Ng's "Deep Learning Specialization" course on Coursera. It was made a few years ago but all of the fundamental concepts are still relevant today.
0 replies
48m
Fei Fei Li and Andrej Karpathy's Stanford CS231N course is also a great intro to the basic of the math from an engineering forward perspective. I'm pretty sure all the materials are online. You build up from the basic components to an image focused CNN.
0 replies
1h23m
Ya there are concepts in programming and math that are mostly self-teachable from first principles, but then there's what looks like gibberish because it's too new to have been distilled down into something tractable yet. I would say that arrays and matrices are straightforward to understand, while tensors are not. So I'm disappointed that so much literature currently revolves around tensors. Same for saying embedding instead of just vector representation, etc.
It helps me to think in terms of levels of abstraction rather than complexity. My education stopped at a 4 year degree, but AI is mostly postgraduate still. So I have to translate to what I know because I haven't internalized the lingo.
Here's the most approachable teaching of neural nets (NNs) and large language models (LLMs) that I've seen so far:
https://news.ycombinator.com/item?id=40213292 (Alice’s Adventures in a differentiable wonderland)
https://arxiv.org/pdf/2404.17625 (pdf)
https://news.ycombinator.com/item?id=40215592 (tensor and NN layer breadcrumbs)
II A strange land 105
7 Convolutional layers 107
..
7.1.3 Translational equivariant layers 112
..
9 Scaling up the models 143
..
9.3 Dropout and normalization 151
9.3.1 Regularization via dropout 152
9.3.2 Batch (and layer) normalization 156
III Down the rabbit-hole 167
10 Transformer models 169
10.1 Introduction 169
10.1.1 Handling long-range and sparse dependencies 170
10.1.2 The attention layer 172
10.1.3 Multi-head attention 174
10.2 Positional embeddings 177
10.2.1 Permutation equivariance of the MHA layer 177
10.2.2 Absolute positional embeddings 179
10.2.3 Relative positional embeddings 182
10.3 Building the transformer model 182
10.3.1 The transformer block and model 182
10.3.2 Class tokens and register tokens 184
11 Transformers in practice 187
11.1 Encoder-decoder transformers 187
11.1.1 Causal multi-head attention 188
11.1.2 Cross-attention 189
11.1.3 The complete encoder-decoder transformer 190
11.2 Computational considerations 191
11.2.1 Time complexity and linear-time transformers 191
11.2.2 Memory complexity and the online softmax 192
11.2.3 The KV cache 194
11.2.4 Transformers for images and audio 194
11.3 Variants of the transformer block 1970 replies
43m
understand but don't really get
That's exactly where I am at. Despite watching Karpathy's tutorial videos, I quickly got lost. My highest level of math education is Calculus 3 which I barely passed. This probably means that I will only ever understand LLMs at a high level.
2 replies
1h31m
Not as a starting point.
Google and find the examples where someone does it in a spreadsheet. It's much more approachable that way.
You are going to find it's not that complicated.
1 replies
1h28m
Sounds interesting. Do you have a link?
0 replies
1h23m
1 replies
2h59m
Only do it if you want the illusion of LLM's to be shattered. Suddenly every day you'll see two to three highly upvoted links on HN and be unable to keep your eyes from rolling.
0 replies
1h33m
that's like saying if you study real neurons your illusion of the human mind will be shattered.
0 replies
32m
This was posted on HN a while ago and led to some great discussion. Myself and others agreed that this type of stateful visualization was _way_ more effective at conceptualizing how an LLM works than reading code or stepping through a debugger.
7 replies
2h41m
I know its not really related but I've noticed something that is making me feel out of touch. Lately there seems to be this increasing merge of tech with weeaboo culture. I may not have the term exactly right but I am talking about the anime girl in the OP's blog post. Its not everywhere but I've started to notice, so it is increasing. Did I miss something? Is this replacing meme's in tech speeches? (I was never fond of that either so I guess I'm a curmudgeon or perhaps my ADHD brain just finds it too distracting)
The post looks informative I hope to learn something from it later tonight. Thx
1 replies
2h28m
It has. I find it infantile and reflective of general millennial peter pan syndrome sensibilities, personally. (i'm a millennial fwiw) But clearly I'm in the minority.
I mean wtf is this. https://kubernetes.io/blog/2024/04/17/kubernetes-v1-30-relea...
0 replies
1h59m
Millennial too. Not to shift blame, but from observation it seems to be more of a gen z thing.
Anime/waifu shit, furries and all becoming commonly accepted as of late? 10-15 years ago you'd be exiled. Now it seems like it's whatever
0 replies
2h23m
I'd say it's nothing more than a generational shift in popular culture... brace yourself for future anime memes.
0 replies
2h6m
its not really related
It's also very much offtopic since it generates repetitive thread-gobbling tangents, like this one is threatening to. Mentioned in the site docs a couple of different ways:
Please don't pick the most provocative thing in an article or post to complain about in the thread. Find something interesting to respond to instead.
Please don't complain about tangential annoyances—e.g. article or website formats, name collisions, or back-button breakage. They're too common to be interesting.
0 replies
2h18m
It's because a lot of the users of gen ai are generating anime waifus. Better gen ai = better waifus. It also helps that devs and programmers are a group that is already likelier to be into anime. Generative AI's killer app is the AI girlfriend / boyfriend.
0 replies
2h15m
It isn't new. In fact, in Tokyo, Japan, Akihabara "electric town" is both the tech mecca and the anime/manga/otaku mecca. Same for Den-Den in Osaka. In the west, the weeaboo movement has always run alongside tech. I guess nerds/geeks and otakus are of the same kind. It does not mean that all tech guys are weebs and all weebs are into tech, but there is definitely some correlation.
Why? I don't know. Video games may be a common denominator. Also, Japan was really big into tech in the 90s, and they still are to a lesser extent.
0 replies
2h31m
I'm still waiting for furry artwork to become culturally acceptable in technical lectures. I briefly snuck a cute Lucario/Zeraora drawing into a presentation on my college final, and the critical reception has been promising, so far.
4 replies
22h29m
I don't want to be dismissive, it's a fun project, but this has been done a lot already - maybe not with llama3 but the architecture is basically the same as llama2. Look at the big list of from scratch implementations on Karpathys llama2.c page.
Is there something particularly different about this one?
Edit - guess not?
0 replies
21h47m
There's literally an image of Anya pointing at Karpathy on this GitHub page.
0 replies
22h9m
Well given the fast pace of AI, it should not be a surprise that this is similar to llama2 and that we’re seeing the n + 1 toy implementations and likely has bugs or leaks in the background.
You might as well look at llama.cpp for a serious and production grade implementation to learn from. Otherwise, nothing to see here.
Is there something particularly different about this one?
Other than the immature lowercase, anime BS, etc, then…
No.
0 replies
22h10m
I think they learned a lot doing this? And they tried hard explaining each step!
0 replies
7h51m
What are your favourite implementations of a GPT? I like a lot the video series by Karpathy.
Anyway, I'll take a look at this too, not sure if it has inference and training. Having just inference would be a disappointment.
2 replies
1h5m
Aaaaaaaaaa.org is possibly the worst domain name I've ever encountered in all my time using the internet. I support your mission but you need to change that.
1 replies
55m
While I agree with you, it's easy to remember using a simple rule. A*10
0 replies
23m
a8a would be the typical numeronym
1 replies
23h2m
Are you the repo author or reposting something cool? I am curious because I want to talk to the repo author about a collaboration project.
0 replies
22h59m
You might be able to reach the repo author on X: https://x.com/naklecha
0 replies
4h51m
amazing work
0 replies
22h5m
I'm sorry but this is absolutely unreadable.
0 replies
22h57m
Genius.
0 replies
8h43m
I'd like to see this using ONNX and streaming from storage (I have my reasons, but mostly about using commodity hardware for "slow" batch processing without a GPU)
0 replies
2h2m
hey, thank you for sharing my project! this made my day <3
0 replies
23h22m
Awesome, gonna go through!
0 replies
4h33m
this is a proper post
0 replies
8h12m
The Spy X Family girl really adds to my enjoyment of this
0 replies
6h36m
She can read your mind llama
0 replies
15h18m
starred
0 replies
18h44m
This is implementation of the inference part and not the training part, right? I’d love to see the training part open sourced and annotated like this.
0 replies
2h30m
dingboard w
0 replies
19h44m
I wanted to try the repo by Karpathy, but I still don't want to learn C (Llama is probably his only C repo), so thanks for posting this.
Interest to know why it is off putting.
Do you need cartoons of children in your readme to get the point across?
Does Docker need this "cartoon" of an otter to get the point across? https://github.com/docker/docs?tab=readme-ov-file
or this "cartoon" of an octopus? https://github.com/docker/compose
This seems to really just be "oldman-yelling-at-clouds-syndrome"
I for one welcome anime girls in readmes and hope to see more of it in the future if only because it seems to bother some of the old hoagies in the world for some reason.
I'm glad you enjoy anime girls but surely you can see why it's different than a project's logo?
One is directly related to the project, the other isn't. It's not even contextually related.
Python (the language) is named after "Monty Python's Flying Circus" simply because Guido was reading the scripts at the time:
Why does github use an octocat as its logo? It's unrelated to software development
The cartoon is literally pointing at contextually relevant information, and it's far more pleasant to follow than yet another big red arrow. That said, I would have enjoyed my reading a bit more if the author utilized a more diverse cast of characters.
Is 29 considered old hoagie?
Old hoagie is more of a mindset. Anyone of any age can be an old hoagie if they like, all one has to do is practice getting upset when one sees anime girls, believe in the coming AI apocalypse and use Emacs.
Don't see how Emacs fits into this. At least I can sort lines there without another proprietary addon.
Does github need a cartoonish cat with 5 octopus-like legs to be its logo? Of course not, but it makes it memorable and funny. And besides, anime is extremely mainstream these days.
Wouldn’t quite go that far. I’ve only met one anime fan in my entire career.
Do you ask everyone you meet?
Then you must be old. Even in western countries Spy x Family (which the character is from) has sold millions of copies, while most people read mangas online and won't be counted. In the country I am from I frequently see people wearing merch of it, mostly because Uniqlo has had a successful line of it. And that is just one manga/anime out of hundreds of popular ones.
Using anime characters is similar to boomer nerds referencing Marvel/DC comics , Star Wars etc.
I would likely be just as put off by a picture of Spongebob or Goofy or Goku in a readme as Anya, fwiw.
maybe you should evaluate whether arbitrary societal norms of "professionalism" or something else are leading to you miss out on cool stuff
I wouldn't have prepared information this way, but judging by the immense popularity of _why in his day, I'm forced to assume that many prefer to have the cartoons
Those cartoon foxes secured his legacy, and to a significant extent, that of Ruby itself.
I would agree putting a cartoon character in readme, without any good context is definitely unprofessional. But would not go as far as offputting.
This statement is simply a variation of an ad hominem attack. It chastises the creator based on appearances that do not align with the niceties that the commenter deems appropriate.
There is a time and place for everything. This isn't it.
In your bubble. In mine this is totally fine, even encouraged.
Indeed. In my company Slack, our primary professional communications tool, I can count a few people with anime avatars. Not very many, but it counts.
yuck
Agreed. For me, the anime character is not "creepy" at all. In fact, I've seen various ML blogs use manga characters to guide the reader.
I have found his lack of proper order, grammar, punctuation, etc... is what lost me out there. This style is fine for 3-4 steps tutorial. But if you have something this long, then you need a proper Table of Contents and make sure to make it a professional old-fashioned doc.
The lack of punctuation and capitalization is a weird zoomer style of writing in lowercase because "it's more chill." It is very common in people < 25 years old. They'll grow out of it.
You get ToC for free with GitHub's README renderer (top-right corner).
She is from a manga / anime called Spy × Family which has 8.3 on iMDb. The best spy on the planet pretends to be a family man for deep cover by adopting the girl (who can read minds, he doesn't know this) and quickly marries a woman (who is an assassin also looking for cover, he doesn't know this). They do their missions in-between roleplaying a perfect family.
https://www.imdb.com/title/tt13706018
I'm OK with that. I did find it distracting, because I knew the character (not very well, I thought the kid was the assassin) and the overall conceptual juxtaposition was... weird.
Beats a cheery AI voice, though.
Maybe it works for a younger generation of nerds? Don't judge a book by its cover.
DbxduuuhhhhÀdcs VC dem s
If this is the case, I feel as if you will be put off by a significant portion of ML engineers.
Security programmers and dev-ops people too. Two areas famously disproportionately represented by furries and co.
It made is 10x better for me. Stop being boring. I like the anime. It's a popular anime. Loads of people like it and think this is funny.
It should be obvious that not liking something does not implying being boring.
If young girls are creepy to you, you should stop watching B-tier horror franchises.
Well that escalated quickly...
Creepy??
I must say I find your comment off putting.
ok boomer
Just treat it as a weird watermark. That's what works for me.
I didn’t not find it off putting. I found it quirky and less boring.
he's using dingboard.com to edit his images. i believe the anime girl is one of the default images (or used to be) on a new canvas.
Do you seriously not find this hilarious?
https://github.com/naklecha/llama3-from-scratch/raw/main/ima...
You should be off pudding
Have you looked at various models on Hugging Face? There are so many anime characters headlining the readme's. I think it's an interesting cultural disconnect to observe in this thread, but at the end of the day, open source projects like this are not obligated to be anything in particular, and entirely subject to the author's tastes.
will not stand this anti-anya slander
I read this comment and I thought you were upset that it was sexualized, but when I looked, it wasn't at all. It might have well been a cute kitten or puppy doing the pointing, hard to get wound up about.
I don't know why this is such a hot take.
Personally, I find it distracting when some devs start to "spice up" their presentation with manga characters, furry characters, memes, or whatever stuff they enjoy.
Shit, I love Zelda - but I wouldn't want Link all over my presentations. It just looks...juvenile and unprofessional. Doesn't mater if you're a beginner or world leading researcher, just keep it simple and undistracting.
EDIT: That said, I'm probably not the intended audience for this piece.
it's not creepy, it's from a popular anime/manga. It's just that the right wing in America (and other western nations) has tried to make us all feel guilty about anime because it doesn't fit their puritanical outlook on the world and that "the other" is bad, evil, and perverted, even though manga/anime has been mainstream for at least 3 decades now. Face it, not all the animation in the world has the same style and look as "traditional" USA animation or comics. Would you have been offended if it was the Charlie Brown kids?
boring...
It's fun. Not everything has to be dry.