HNHacker News
TopNewBestAskShowJobs

rollinDyno

494 karma · joined August 28, 2015

hi@faure.hu
submissionscomments
rollinDyno··on Ask HN: What words do you mute on Twitter / X?
I built the list while scrolling on my feed so if you knew me it would make sense that I haven't felt the need to mute tankies.
rollinDyno··on Ask HN: What words do you mute on Twitter / X?
I asked because I'm building my own mute list. I realized that I'm spending too much time consuming these news on Twitter and I'd rather stay informed from better quality sources.

The list I have come up with so far is: @AOC, AOC, @KamalaHarris, Harris, Buttigieg, Milei, Putin, @nytimes, Headline, Washington, Governor, Israel, Iran, Conservative, Liberal, New York Times, Kennedy, Socialism, Communism, Musk, Elon, Biden, Obama, Clinton, Primary, GOP, Republican, Democrat, Tankie, Libertarian, Poll, Walz, Kamala, Vance, Trump.

If you are taking any terms from my list (or are already muting them), please do not completely tune out of US politics. It is very valid to decide that the cultural wars going on Twitter/X are purely noise, but it is a big mistake to think that because most loud people can only engage on a gaseous level, then public policies (and by consequence elections) do not matter and are not relevant in your lives.

rollinDyno··on Visualizing Attention, a Transformer's Heart [video]
If the point was the presentation of only the next token, and predicted tokens were a function of only the preceding token, then the vector of the new token wouldn’t have the information to produce new tokens that kept telling the novel.
rollinDyno··on Visualizing Attention, a Transformer's Heart [video]
That weighted context is the 12228 dimensional vector, no?

I suppose that when you each element in the vector weighs 16 bits then the space is immense and capable to have a novel in a point.

rollinDyno··on Visualizing Attention, a Transformer's Heart [video]
Hold on, every predicted token is only a function of the previous token? I must have something wrong. This would mean that within the embedding of "was", which is of length 12,228 in this example. Is it really possible that this space is so rich as to have a single point in it encapsulate a whole novel?
rollinDyno··on Show HN: Anki/Duolingo-like app using educational YouTube videos
When I say I compile information I mean I have a scratchpad in which I fill in the evidence that is relevant to my current needs. So I'll have the source noted there.

What I'm trying to say is that I rarely go back to the scratchpad of a previous question / problem, but I will approach every question and problem with a new search.

Trying to save information for the sake of it without knowing what future use it is going to have stresses me out, so I find that this fresh scratchpad approach works best for me.

rollinDyno··on Show HN: Anki/Duolingo-like app using educational YouTube videos
I don't even save the sources, I'll just assume I'll always have access to the internet and that I'll be able to find what I need when I'm thinking about it.
rollinDyno··on Show HN: Anki/Duolingo-like app using educational YouTube videos
It depends on the question. I do not have a list of sources.
rollinDyno··on Show HN: Anki/Duolingo-like app using educational YouTube videos
I've always wanted to do this for Wikipedia, it could even be a Wikimedia add-on.

However, I have recently transitioned towards becoming better at compiling information quickly rather than spending a chunk of my day memorizing facts that I am not quite sure will be useful.

rollinDyno··on Lets Have Three Abstraction Layers
I find it interesting that as our societies have become increasingly complex and we have organically implemented what we think is an efficient division of labor, we have ended up with people not only specializing across multiple sectors, but also in different layers of abstraction within the same sector. We have done all of this without even realizing that we were pushing people towards a different interface of the problem, but the same problem nonetheless.
rollinDyno··on The new science of meditation
What reasons do you have to trust your intuition? Why do you happen to have a certain intuition and why is it likely to give you the right answers?
rollinDyno··on So you want to learn physics (2021)
This post comes in at a very good time because I have recently begun to become interested in particle physics and have so far only resorted to watching YouTube videos. This is a sign that it is maybe time for me to jump into a textbook.

However, I seem to be interested in a few particular questions about particle physics as a science rather than facts about particle physics. For instance, I am interested in the instruments and methods that physicists use to verify their theoretical claims empirically. I am also interested in how theorists are able to come up with theories so early on, such that they are confirmed by evidence many years later. What are the assumptions that they were able to make? I am curious about where they derived the creativity to be able to bring in so many assumptions together and the come up with their models. Now that I write this, I realize that before theories were validated there were probably competing models.

Therefore, I am not exactly sure I want to study particle physics per se, or whether a book on the history of particle physics will do. I am ok with having a popular understanding of the subject, I mostly want to gain inspiration from following the work of famous scientists.

rollinDyno··on Sens. Warren and Graham to Introduce Digital Consumer Protection Commission Act
From the Warren and Graham opinion piece on NYT:

> Our Digital Consumer Protection Commission Act would create an independent, bipartisan regulator charged with licensing and policing the nation’s biggest tech companies — like Meta, Google and Amazon — to prevent online harm, promote free speech and competition, guard Americans’ privacy and protect national security. The new watchdog would focus on the unique threats posed by tech giants while strengthening the tools available to the federal agencies and state attorneys general who have authority to regulate Big Tech.

I'd like to see the actual language of the bill. For now, in these public-facing articles, it seems as if the scope is too borad. For instance, what is 'Big Tech'? I'm also concerned about an agency responsible for 'unique threats'. While it's loable that there'll be an agency in charge of addressing the universe of threats that are not within the purview of any other government agency, this could be easily abused.

rollinDyno··on Ask HN: Is There a Crunchbase for Policitians?
You can always request this data directly from the SEC or the FEC for the raw reports. In my experience, the regulation is does not provide enough transparency. The three main issues I can think of right now are the following:

- Citizen's United. No need to elaborate.

- It is incredibly hard to marry lobbying and fundraising data. The company names are not the cleanest.

- Extent of enforced disclosure in the Lobbying Disclosure Act: lobbyists do not need to say which bill they are trying to influence, or which official (political or bureacratic) they are trying to influence / contact.

Both acts have tiny loopholes. I don't remember the exact quantities, but if you're donating an amount smaller than X to a politician, it does not need to be reported. And if less than Y% of your time is spent attempting to influence politicians / government agents, you don't need to report this.

rollinDyno··on Meta Reports Second Quarter 2023 Results [pdf]
I have a country level variable and I want to run a basic regression.
rollinDyno··on Meta Reports Second Quarter 2023 Results [pdf]
I wish per country data was available. I’m currently trying to estimate Facebook’s ARPU per country.
rollinDyno··on How to stay safe and happy on an electric unicycle
I currently use Citibike and am happy with it. However, if I ever considered purchasing an EUC instead of a bike it's because I wouldn't have to park it and thus making my trips door-to-door rather than having to stop by wherever the bike is parked. I would be able to bring it with me, whether I'm going upstairs into my office or grocery shopping (as long as they have a collapsible handle.)

I think I am more persuaded by the EUC than by the electric scooter because of its speed and maneuverability? I'm not sure, but it seems to have similar problems as the bike.

rollinDyno··on Electric bike, stupid love of my life
I am considering buying an e-bike. However, I live in NYC and have access to Citibikes. I am a member so I can ride regular bikes for free and pay pennies for short trips on last-gen bikes.

So if I pay $1000 for a Roadster V2, what exactly am I getting? Sure, Citibike can be inconvenient: sometimes I won't find a good bike and sometimes I won't be able to park. For this, I'm not sure it's worth paying the prize.

Having my own bike also exposes me to risk of theft and having to carry it in and out of my apartment (I don't currently have bike storage in my building). Hence, it seems like I've already found my optimal solution.

rollinDyno··on The Problem with LangChain
People have already said that LangChain is useful to get ideas about what can be done with LLMs past the single prompt-response dynamic. Something else I thought to add is that LangChain is also useful for prototyping, so you'll be able to have a proof of concept of your idea before the weekend ends. I do agree that once you have that nailed down then you should rewrite everything from the ground up.

I do recommend prototyping; even if you think your idea is feasible, you want to see it in action before committing significant resources. Personally, my prototype showed me that a single run of my idea takes more time and money than I initially expected.

rollinDyno··on Proof you can do hard things
I raised the same questions as Nat, but I think he hasn't really discovered why we were asking that. The reason why I was looking for a good reason to study calculus is because I was lazy, and if I'm lazy enough then I can change the definition of what "hard" means. For instance, I could convince myself that if my could teach calculus, and him being just another dude, then surely I could learn calculus.

I would also be cautious about setting yourself up for a hard life. The takeaway resonates well with the HN crowd, myself included, because we like challenging ourselves. There's a lot of people out there who are simply looking to satisfice their lives, and you'll need another way to motivate them to learn calculus.

rollinDyno··on LlamaIndex: Unleash the power of LLMs over your data
I gave this a shot a while back and found plenty of examples but little documentation.

For instance, there is a tree structure for storing the embeddings and the library is able to construct it with a single line. However, I couldn’t find an clear explanation of how that tree is constructed and how to take advantage of it.

rollinDyno··on JupyterLab 4.0
I see. If I faced the same constraint I would try to find whether VSCode can access files remotely, which is very likely.
rollinDyno··on JupyterLab 4.0
I have been using VSCode notebooks with .ipynb file extensions, this gives me many advantages as I am able to configure things I'm not able in JupyterLab. I also have access to a very rich ecosystem of plugins. If there is anyone aware of VSCode as a solution but keeps using JupyterLab, could they explain why?
rollinDyno··on Average color of the NYC sky every 5 minutes
Very cute, although I'd like to see what the results are like when choosing the dominant rather than the average color.
rollinDyno··on Google DeepMind
When was the last time Google was proactive rather than reactive? It feels that this is the same for all big Tech firms except for Microsoft.
rollinDyno··on ChatPDF – Chat with Any PDF
Unfortunately, this is not ready for the sort of papers that I read. I mostly read papers with regression tables or papers with closed-form equations.

I have tried using Mathpix to convert the formal theory papers into latex and then fed it to GPT-4, but it was not able to take the whole text in a single prompt. When I broke it down in multiple prompts, it started responding hallucinated sections of the paper. I had given preemptive instructions stating that I was going to share the paper section by section and then ask it questions.

Once I finished uploading the whole paper after multiple prompts, it did not give satisfactory answers.

rollinDyno··on Show HN: IPython-GPT, a Jupyter/IPython Interface to Chat GPT
Does anyone know if I can integrate this on Quarto? I use VSCode.
rollinDyno··on Using LLaMA with M1 Mac and Python 3.11
I've been commuting for about 45 minutes on the subway and I sometimes try to get work done in there. It'd be useful to be able to get answers while offline.
rollinDyno··on How to Annotate Everything
I am a strong believer in taking-notes so that I don't have to face a blank canvas every time I want to start a new essay.

My issue is that when I am in "the zone", I can't write. When I am very intensely focused on a topic, I am informing my internal conversation very efficiently by skimming papers. If I interrupt that with taking notes, then I am adding unnecessary friction that slows me down.

When I write notes, I feel that my ideas race ahead of my typing speed, and my working memory is not large enough to keep these ideas in a buffer. This is an issue I have had my whole life but only recently noticed it is an impediment and only now can I describe it.

All that being said, even if this wasn't an issue, there's a trade-off between how quickly you move across text and how much of it you store on notes. There's an optimal point, of course, and this can be raised with technology, focus, and practice.

rollinDyno··on Unbundling Tools for Thought
I am a strong believer in taking-notes so that I don't have to face a blank canvas every time I want to start a new essay.

My issue is that when I am in "the zone", I can't write. When I am very intensely focused on a topic, I am informing my internal conversation very efficiently by skimming papers. If I interrupt that with taking notes, then I am adding unnecessary friction that slows me down.

When I write notes, I feel that my ideas race ahead of my typing speed, and my working memory is not large enough to keep these ideas in a buffer. This is an issue I have had my whole life but only recently noticed it is an impediment and only now can I describe it.

All that being said, even if this wasn't an issue, there's a trade-off between how quickly you move across text and how much of it you store on notes. There's an optimal point, of course, and this can be raised with technology, focus, and practice.

← PreviousPage 2 of 5Next →