daquisu

It was a recent edit though. Yesterday snapshot: https://web.archive.org/web/20260613072958/https://huggingfa...

Later in the same blog post, the author says:

We can also consider the IMO 2025 problems individually. In the Epoch AI newsletter, Greg Burnham combines a subjective analysis with Evan Chen’s MOHS ratings to argue that the first five problems at IMO 2025 were unusually easy and the sixth was unusually hard, so it’s not surprising that the first five problems were exactly the ones solved by these AIs. Though I’m not sure the MOHS scale is rigorous enough to make sense as the x-axis of a bar chart it’s easy to corroborate the high-level story with the official IMO statistics. Based on average scores, this year’s Problem 6 was the fourth hardest and its Problem 3 was by far the easiest of all Problem 3s and 6s since 2000.

In the linked MaxProof paper, in the section "6.3.1. Per-Problem Analysis" it shows the same behavior: 7/7 in the first 5 problems, 0/7 in the last problem.

"I thought it was interesting and a bit underappreciated that the fraction of gold medalists at the 2025 IMO (72/630 = 11.4%) is the highest it’s been since 1981.

Crudely, IMO gold medals are awarded to the highest-scoring 1/12 of contestants.1 However, because scores are integers up to 42 and there’s no provision for tiebreaking, it’s possible for a lot of contestants to be tied around the threshold. In that case, either all of them get a gold medal or none do, and the fraction of gold medalists might deviate substantially from 1/12. That’s what happened this year: 46 contestants all won a gold medal by scoring exactly 35 points.

In fact, bizarrely, 35 is the mode of the scores this year; the last time the modal score was a gold medal score was in 1994. And, of course, 35 is the same score claimed by AI systems from Google, OpenAI, and others."

From https://blog.vero.site/post/imo-2025

I just tried that since I read your comment and it is really helping me. Thanks!

Now it is even easier. Cloudflare has a beta product called AI Search that implements most of these 160 lines of code

12.5 million a year for a hundred people seems reasonable? 125k per person per year. GP still said "a few hundred" - two hundred would drop that value to 62.5k per person

Firefox on mobile works with ublock. It can also play videos even with the screen locked, although you do have to unpause it after locking the screen.

The "You are an expert software engineer" really helps?

Anecdata, but it weirdly helped me. Seemed BS for me until I tried.

Maybe because good code is contextual? Sample codes to explain concepts may be simpler than a production ready code. The model may have the capability to do both but can't properly disguished the correct thing to do.

I don't know.

That is a common narrative but Google had LaMDA as an LLM with over 100B parameters before the ChatGPT release. There was even a Xoogler that claimed it was alive.

From my POV Google could have released a good B2C LLM before OpenAI, but it would compete with their own Ads business.

Which better measurement do you propose?

It is done by the extension without any fancy stuff. Extensions can load static js / css and bypass CSP with it, if it is declared in their manifest.json. Grammarly's manifest.json is here: https://gist.github.com/Daquisu/11eb1a7000b4141c4404edcc6e16...

For more advanced CSP bypass with extension, you can:

1. Inject JS code into any webpage with a CSP.

2. Create an event listener for your content script and reacting according to it.

3. Use your content script to communicate with the background script.

4. Use the background script to communicate with any website, including blocked websites by the CSP.

Basically, any website <-> extension content script <-> background script <-> any website.

Weird, they released Gemini 2.5 but I still can't use 2.0 pro with a reasonable rate limit (5 RPM currently).

See: the parameter "temperature" for LLMs

It is interesting how France became so focused on analysis and properly proving theorems and stuff, while the applications don't have the same highlight in prépa.

One professor of mine commented that most French engineers are better mathematicians than most mathematicians in Brazil.

It is the opposite of what the linked article mentions that was happening in Weierstrass' time.

Regarding a laptop with a removable wireless keyboard, ZenBook Duo has that, although the touchpad is removed with the keyboard.

It also has two screens and its own stand, I use it as my travel machine.

https://archive.is/UlZUR

This seems easier to get right than if( x = *p++ )

For people with native or fluent English, for sure. For the others, probably not.

Hey OP, thanks for the text. Posting some provocations here since I like to expose an introverted point of view.

When everyone else can, why can’t you just enjoy a drink or two, and be part of the group for a few hours?

Apart from the hearing loss, maybe you are an introvert and just prefer more quiet enviroments. Nothing is wrong with that, there is no need to feel you must go out because of peer pressure, and no need to wait until you have a hearing loss to have an excuse to stop going. It is better to find other people who are introverts than eternally trying to adapt to be something you may not want.

At some point it stops making sense, having done it repeatedly with the same outcome.

If you find purpose in doing something, then you don't do it because of different outcomes. Maybe you didn't find purporse in going out from the beginning?

What you are describing seems like a list of pull requests for a repo you are familiar with.

(Ideally) short code snippets, discussion about what it solves, reviews possibly with other alternatives, discussions about tradeoffs...

To be fair, there is some research about IQ having a genetic component, which one can do by comparing twins with the same DNA and twins without the same DNA [citation needed]. People who are curious and have the drive to keep learning are going to have a higher IQ, and personality has a genetic trait.

Just like you, I think IQ is very trainable. It is very common to have questions about logic, which may be represented as Venn Diagrams, or visual questions with multiple parts, each one following a logic...

<appeal to authority> I say that as someone who "started" with 110 QI according to a Orkut online test, and grinded my way to 145 according to different online tests. </appeal to authority>

For that reason people who learned how to prove things in math are naturally going to have a higher IQ: they learned how to solve some questions in the test.

I played quite a lot of league of legends (probably 10k games), but I stopped playing all compute games in the last two years.

Competitive games in general feel like STEM olympiads: you have an objective ranking, you have challenges, you have competitition. You deal with hard problems related to performance: how to keep improving? Which topics are hard for me?

It was crazily important for my personal development, but as with olympiads, eventually there isn't a lot of new things to learn, at least things that you will carry with you for life.

That's the joke. You can scroll down below to see the reviews talking about how unproductive they are while using the extension.

But good you at least opened the link and raised the question, there a lot of people here supposing it is a blocker to be more productive.

I realised recently China has a population of 1 billion and 400 million. That's more than a billion more people than the U.S.

That's funny because India has also over a billion more people than the U.S., but the U.S is ranked 3rd by population. This means that even if US got 1 billion more people, they would still be the ranked 3rd by population

Perhaps myopia turns into astigmatism if you squint too much?

In my case no. I started with -2.5 for myopia, and was squinting pretty hard for a few months before that. Had no astigmatism

Then I got a bad habit of scratching my eyes whenever I slept too little time. While this happened my astigmatism increased by 0.25 D each year, for 4 years in a row