i'm not an expert but i think you might be conflating uva and uvb. uvb is the one that goes to zero in winter and near sunrise and sunset, is completely blocked by glass, and requires being near solar noon to have maximum effect (i.e. highest when your shadow is shorter than you). It is also the one that is responsible for vitamin D synthesis.
on the other hand, uva is the one that is still substantial in winter and away from solar noon, goes through glass easily, and can cause sunburns and skin cancer.
I genuinely wonder how this style minimized the loss function or got the most upvotes in RLHF and yet is so universally hated that it gets flagged to death almost every time, and similar to Reddit. If I were to describe it, it's "snappy" and information-dense, without fillers. I dislike it too of course.
This style is deliberately trained by the frontier providers, not something that occurs accidentally. It is a manipulative style that is extremely effective against the general population. It utilises countless dopamine-inducing techniques used in clickbait headlines and Youtube thumbnails, and barrages the user with a wall of text that obfuscates everything it attempts to say, which is extremely useful for giving the appearance of intelligence; when you use a lot of sophisticated language and technical jargon, people won't understand you, but rather than assuming that you're stupid for writing something incomprehensible, they will instead give you credit and assume they can't comprehend it because it's too advanced for them, even if actually is incoherent.
It is only flagged on HN because it's been made against the rules, giving the minority who hate it the power to retaliate. It was only some months ago I was routinely getting downvoted every time I pointed out obvious bot accounts spamming a post per minute in blatant LLM-speak. Even now, LLM articles are still allowed and people upvote them to the top all the time.
> information-dense
Err, no. It absolutely is not. You could say that it's dense in technical language, but the style has mastered the art of saying a lot without saying anything at all.
Your comment sparked an idea - it’s like the bike shedding of written text. If you write clean, concise English, it’s easy to parse and then inject your thoughts. If you instead (like an llm) throw a human a large wall of jargon heavy text, they’ll just give up and agree with you.
reduce has complexity to handle the edge case of an empty iterable, and also for the case of a binary function with different types for inputs and outputs. That makes it harder to reason about and "uglier" than map and filter. People probably hate sum and product significantly less, both those also have the edge cases of empty iterable, in which case the natural result is 0 for sum and 1 for product sure, but of what type?
While we're on the subject, can someone explain to me why in Rust, you need to annotate the type when you call .sum() on an iterable? For example
let p: i32 = [1i32, 2, 3].iter().sum();
println!("hello {}", p);
That works, but fails if I replace `p: i32` with `p` or `p: i64`, and I cannot find a satisfactory answer in any thread or llm. The obvious question is why the compiler cannot infer the type from the element type of the container, and the naive response to that is for flexibility summing into a bigger type. But in that case, why would `p: i64` be rejected? And what other type is allowed besides i32?
State-sponsored psyop meta comments aside, the models obviously continue to get better, but there is still a lot of 'guard railing' required to keep even the latest models completely on-task. The chess example is interesting because it's clearly a well-studied and established domain so the rules, strategies, and whatever else is in the training data should make yield excellent results; but clearly there is some behavior in these systems that's difficult to engineer out.
I'm not sure why anyone is expecting stochastic systems to be deterministic.
Chess is a deterministic game won by a combination of known movesets and constrained multi-level forward search.
LLMs do neither of these things. They don't reproduce training data exactly, their next response is more 'inspired by' prompts and its own memory than produced deterministically, and they don't have the capability to do general forward search on their own.
So when you ask an LLM to play chess you're getting the equivalent of a very compressed and lossy JPEG of chess rules and strategies with added per-turn random noise.
They also don't have the ability to design their own chess engine, although it would be interesting to see what happens if you ask for one.
For me the useful intuition is that LLMs haven't somehow magickally learned to implement any of the algorithms we know that we have used to make strong chess engines: alpha-beta minimax and Monte-Carlo Tree Search on the one hand, and obviously the ability to learn accurate evaluation functions by self-play.
I mean we've done all this before in a task-specific fashion. It's useful to know that LLMs haven't managed to do that in the process of learning to represent the entire text on the web. On the other hand they have gotten say very good at machine translation without being trained exclusively (and I select the preceding word carefully) on machine translation.
Edit: I'm saying this because there is this idea expressed by e.g. Ilya Sutskever, that in order to predict the next token accurately an LLM has to learn something about all of underlying reality. See for example this interview with Dwarkesh:
Where Sutskever claims that "Predicting the next token well means you understand the underlying reality that led to the creation of that token".
If that were true, we should have seen LLMs play good chess by now. There is a huge amount of data on playing chess floating around on the web in the form of algebraic chess notation and if LLMs were capable of learning the "underlying reality" of chess, they would already have. They haven't. Because they can't. What Sutskever is saying flies in the face of literally hundreds of years of statistical modelling, which is to say, building predictive models that, very explicitly, do not have to understand any "underlying reality" and only have to be good at modelling a dataset.
>If that were true, we should have seen LLMs play good chess by now.
Not at all. LLMs learn by imbibing a mass of relationships as isolated fragments of information. There is a certain amount of sorting and indexing that happens during the training phase. There is also a certain amount of compute executed on these relationships during inference. LLMs can model processes that fit within the compute budget. Language translation works well because language is lookup-heavy while being light on compute.
Chess is a compute heavy game of finding the best move out of many possibilities with wide variation in the quality of each move. Humans cut through the compute requirements by reinforcement and learning intuition. LLMs don't get reinforcement on chess so they must compute during inference a unified model of chess. Developing a strong model of chess from raw fragments of information is simply not in their compute budget.
>> Not at all. LLMs learn by imbibing a mass of relationships as isolated fragments of information.
You gotta be careful how you use the word "relation" here because there's an informal meaning (I'm related to my cousin) and a more strict, formal meaning, that is used in computer science e.g. in the "Relational Calculus" etc. In the formal sense, the one relation that LLMs learn during training is the co-occurrence of tokens in a corpus of text, what's called more technically a "collocation" relation. Nothing says that this is enough to play chess, so I'm indeed doubtful that they can.
But they have "learned to implement any of the algorithms we know that we have used to make strong chess engines". Ask Claude Code to write you a chess engine. Your objection is that they don't implement MCTS in the neurons themselves? Neither does a human, we use a C compiler when we want to play chess using MCTS.
That's a separate question from whether an LLM (unaided by a C complier) can learn to play chess as well as human (also unaided by a C compiler). Certainly humans can't become grandmasters only by reading chess transcripts on the web, and certainly humans require many "thinking tokens" during a game to play effectively. Do you know for sure that a transformer can't reach grandmaster level if it is allowed to learn by playing games (as humans do) and is given a sufficient number of thinking tokens during the game? It seems near certain that they could, if someone wanted to spend the money (and I don't see why anyone would.)
Sutskever's claim is that in order to predict the next token a system must learn something about the "underying reality" that produced the token. In the context of chess that means that the LLM must learn something about playing chess (since tokens are the moves in a game of chess). My argument is that contrary to what should be expected if we take what Sutskever says to be true, they don't seem to have.
Yes, I do mean that the LLM's weights are set so that it will execute minimax or MCTS when it needs to. That has nothing to do with whether humans can do the same or not.
I don't disagree that a Transformer could learn to play chess if it was explicitly trained to do that. My argument is that LLMs, trained to predict the next token, have not learned to play chess. That's LLMs, not Transformers.
Just to make sure this is not taken as splitting hairs, the point is that there's all sorts of claims made about what LLMs learn when they train on text. For example, there was a claim by Sundar Pichai that one of their models had learned to translate Bengali without explicitly being trained to do so. It later emerged that Bengali was indeed included in the model's training set [1]. It's not clear whether that included parallel texts, e.g. between Begnali and English or another intermediary language, in any case Sundar Pichai's claim was that the ability to translate Bengali was "emergent".
So I'm interested in understanding the extent to which these "emergent" abilities are real or not. With chess, given the amount of textual data tracing games that floats about on the open internet, I would totally except some ability to play chess to "emerge". Maybe the reported 700-800 ELO level is even that sort of ability. Maybe we should only expect LLMs to learn to play at the level of an untrained, casual player. Maybe not. I have no idea.
On the other hand, the fact they keep making elementary mistakes like illegal moves must be taken to mean that, so far, LLMs haven't learned to play chess.
But it speaks in words, therefore it must be super duper extra smart!!11 /s
Sarcasm aside, I think this is an easy cognitive trap to fall into. It does sometimes feel like the LLM must have some world model because it converses somewhat coherently. Examples like this failure to understand chess, or to count the number of Rs in "strawberry", seem difficult to explain if the models are intelligent. But that doesn't stop people believing they are anyway. I think there must be something about the conversational interface that fools us easily. I wonder if people trained in interrogation techniques are also fooled?
It's a panic in Go, not an exception. In practice that's usually a whole process crash. You can catch panics and kinda use them like exceptions, but it's not conventional.
Oh, net/http recovers from a panic inside a handler and sends error 500. I don't know if that was always the default. Where I worked before, our own Go servers didn't recover, and it did cause outages.
This is kinda bad practice. Go and Rust panics are similar, they aren't exceptions that you can safely recover from. Say there's a panic in the middle of modifying some global state like a database connection. Hence complaints about the net/http recover like https://github.com/golang/go/issues/25245 . But I guess they decided it's worse to let servers crash from trivial mistakes.
reply