Yeah, I would really like if people who introduce a new project to an already very crowded space would start the introduction with "Why MyCoolProject instead of X?" section.
759 karma · joined September 3, 2025
No, it's not. You can still contact the author and ask them to transfer the name to you.
When I need a notebook, I just go into a store that sells them and grab whatever they have. I have a notebook for logging my exercises and another one for random stuff, I have no idea what brand they are.
Moleskin sells fashion items. Or, maybe, an idea of a certain lifestyle. I'm not sure.
Oh yeah, I'm sure you will find lots of competent people. Like Jia Tan, for example. I've heard he is very competent.
> but they also have not credited an artist or provided any proof of human creation
What kind of "evidence" or "proof" would satisfy the poster?
In fact, it is purely additive process: we iteratively add the most frequent pairs to the set, until we reach the desired total number of tokens. But we never remove tokens, we keep everything, including the initial 256 tokens, representing bytes.
This ensures that the model is capable of producing every possible unicode sequence (in fact, I think that it is capable of producing every possible byte sequence, but bytes that are not valid unicode are filtered during sampling).
Edit #1: also, this page entirely skips the attention mechanism, which is, in my opinion, both the most interesting part and the part that is hardest to understand (I can't say that I fully understand it, to me it is just some linear algebra matrix multiplication magic).
If often identifies some of the points of interest correctly, but their spacial location relative to each other is completely insane. Like, not even close to reality.
We had machines "beating" humans in physical tasks for a very long time. No one would be impressed by a car winning a running competition or a construction crane lifting more weight than an Olympic weightlifting champion.
> it's critical to know whether it's vibe coded
Strictly speaking, the only way to be sure that something is not vibe-coded is to either have a proof that the code were published before vibe-coding tools were available or to hand code it yourself.
Also, if you think that knowing if something is vibe-coded is so important, it is unwise to attack people who honestly tell you that something is vibe-coded.
The problem is that your link doesn't really backs up your claims. #1 and #5 deal exclusively with people with various kidney conditions. #2, #3, #4 and #6 deal with issues that are only tangentially related to consuming too much protein.
All research that claims that eating too much protein is harmful is either about people with kidney disease or explores really far-fetched theoretical scenarios.
While harms of consuming too little protein are obvious and self-evident. Every tissue of your body constantly regenerates itself. Generating new tissue is impossible without protein, because protein is what it is made of.
> your unbacked claim
Sure, let's have some links: https://www.uclahealth.org/news/article/are-you-getting-enou...
> “We find most adults are not quite getting enough protein,” says Michael Garcia, MD, a UCLA Health clinical nutritionist.
> When you eat that protein also matters — the protein your body needs must be spread throughout the day. “We’re able to store certain nutrients, but we can't do the same thing with protein,” Dr. Garcia says. “And our bodies can only use so much protein in a sitting and a day.”
> “The recommended amount is really the absolute minimum we need to not fall into a deficient state.”
According to a link[0] provided by another commenter[1] in another subthread of this thread, maximum protein intake is about 2.5 times higher than the recommended protein intake (which is really just "the absolute minimum we need to not fall into a deficient state").
And since most people, who don't consciously control their diets, are very likely to eat closer to the lower bound, I think that telling people to "just eat more protein" is more likely to bring them health benefits than telling people to "just eat less protein"
Which is what TFA effectively does: it tells people about "underrated benefits" of replacing one meal a day by a meal that barely has any protein, which is equivalent to telling people to "just eat less protein".
I think that this article is potentially harmful to people who don't know any better, and I'm very surprised that the issue of reducing protein intake is neither addressed by TFA nor by other commenters. And that my attempt at addressing it receives so much pushback.
[0] https://www.health.harvard.edu/diet-and-nutrition/when-it-co...
> However, for the average healthy person (who is not an elite athlete or heavily involved in body building) it's probably best to keep total protein intake to no more than 2 gm/kg of ideal body weight
So, you can increase your protein intake about 2.5 times above RDA without any issues? I would call this "basically no downsides".
#1 literally says
> For people with healthy kidneys, higher protein intake is generally safe.
#2 addresses the issues of some specific diets ("very-low-carb or ketogenic diets") that "may be low in fiber-rich carbohydrates". I did not advocate for "very-low-carb or ketogenic diets". Also, it doesn't say anything about potential harms of protein itself.
#3 "It Can Crowd Out Other Nutrients"
Again, no mention of supposed "harms of too much protein", only harms of "too little everything else". I never suggested to go crazy and stop eating other foods.
#4 deals with "type of protein" that "may be just as important as the amount of protein" and is irrelevant to your argument.
#5
> A very high-protein diet, particularly one high in animal foods, can increase the risk of kidney stones, especially for people with a history of these conditions or those who don’t drink enough water.
Well, just drink enough water then. Also, I was talking about generally healthy people. I am not qualified to discuss diet choices of people "with a history of these conditions".
#6 "Protein Bars and Powders Can Backfire"
I never suggested any of this. Also, this doesn't address supposed harms of protein itself, so it is irrelevant again.
Of course, you can carefully design other meals to be more protein-heavy, just because of your weird idea of a having an (almost)no-protein meal.
But it is much easier to make it a habit to include at least one high protein food in every meal.
I never said that oats are unhealthy or that carbs are unhealthy. I said that a meal without protein is not the best idea.
My critique was not directed at a particular food item but at the meal composition as a whole.
> I get protein in lunch and dinner, and I don't need my oatmeal breakfast to be protein heavy.
TFA suggests to eat oatmeal for lunch.
> Oatmeal for lunch doesn't make much sense in isolation.
I agree!
I guess, it is possible to make oatmeal for lunch work if you up the protein and somewhat reduce carbs in breakfast and dinner, but way? I think that it is much easier to just make a habit to include a high protein food item in every meal. There is basically no downsides to eating too much protein and there are a lot of potential problems if you eat too little.
> You can enrich it with bananas or goji berries for extra nutrients.
"extra nutrients" are more carbs.
> it’s healthy
I don't think that a whole meal without a good protein source can be considered "healthy"
They are not limiting usage through the API with per token payment.
They are limiting usage through a subscription that gives you a lot (but they won't tell you how much exactly) of tokens for a fixed monthly payment.
Using models through a subscription tends to be much cheaper than using the API and paying per token.
Anthropic provides cheap subscriptions with less profit or even at a loss (we don't know) to promote its own tools. Using subscription for third-party tools obviously defeats this purpose.
What does it even mean?
Claude Code is a so called "harness" - a thing that builds a context for LLMs, calls LLMs, executes tool calls etc. It uses various Anthropic models under the hood.
It can also use other models AFAIK.
It cannot be "trained".
Sorry if this comment sounds nitpicky, I'm just annoyed by the imprecise use of terminology.
I think that, in the early days of internet search, entering full questions actually produced worse results than just a bunch of keywords or short phrases.
So it was a sign of a "noob", rather than a mark of sophistication and literacy.
I find LLM slop much harder to read than normal human text.
I can't really explain it, it's just a feeling.
The feeling that it draaaags and draaaaaags and keeeeeps going on and on and on before getting to the point, and by the time I'm done with all the "fluff", I don't care what is the text about anymore, I just want to lay down and rest.
This is not how the feature called "reasoning" work in current models.
"reasoning" simply let's the model output and then consume some "thinking" tokens before generating the actual output.
All the "fluff" tokens in the output have absolutely nothing to do with "reasoning".
If you launch it in YOLO mode in a separate branch in a separate worktree (or, preferably, in total isolation), you can instead spend time reviewing changes from previous tasks or refining requirements for new tasks.
I mean, at some point it was true.
I remember that around 2023, when I first encountered colleagues trying to use ChatGPT for coding, I thought "by the time you are done with your back-and-forth to correct all the errors, I would have already written this code manually".
That was true then, but not anymore.
It would be better if an LLM coding harness just helped you set up a proper sandbox for itself (containers, VMs etc.) and then run inside the isolated environment unconstrained.
In setup mode, the only tool accessible to the agent should be running shell scripts, and each script should be reviewed before running.
Inside an isolated environment, there should be no permission system at all.