mg

24 ms·

301.
by mg 2y ago
Is there a free (ad supported?) online tool without login that reads text that you paste into it? I often would like to listen to a blog post instead of reading it, but haven't found an easy, quick solution yet. I tried piping text thr
302.
by mg 2y ago
I wonder what percentage of medical issues we could tackle by better analyzing and categorizing the existing medical studies that are out there. And then attaching a user interface to the data that guides the user through a step-by-step sel
303.
Tesla Cybertruck is the third best-selling EV in America(techcrunch.com)
5 points by mg 2y ago|1 comments
304.
by mg 2y ago
I run a self-learning music discovery engine called Gnoosic: https://www.gnoosic.com I can confirm that when you suggest a random band to a random user, they will dislike it with over 90% probability. I'd be interested to h
305.
by mg 2y ago
Hacker News :) It was this Show HN: https://news.ycombinator.com/item?id=7465980 That was before showing something on HN was called "Show HN" though.
306.
by mg 2y ago
I like to use standalone tools with no dependencies, so I made this open source html editor with instant preview as a single html page: https://github.com/no-gravity/html_editor It has a few convenience functions alrea
307.
by mg 2y ago
It sounds like this would make Android devices more useful for developers, which is a good thing. Although I am pretty happy with building what I need via web apps these days. The one thing I am missing on Android is that Chrome on Android
308.
by mg 2y ago
Well, nearly all of the Google Images results for "Woman" show a woman with makeup and additionally the photos were altered via Photoshop. We have been creating our own reality even before AI.
309.
by mg 2y ago
Hmm.. I think this is a misunderstanding. What I meant is: While I am already inside a container running Debian, can I ... 1: ./myscript.py 2: some_magic_command 3: ./myscript.py So 1 runs it under 3.11 (which cam
310.
by mg 2y ago
When I'm in a docker container using the Python 3 version that comes with Debian - is there an easy way to swap it out for this version so I can test how my software behaves under 3.13?
311.
by mg 2y ago
Nice one! Some observations: This fork throws errors in my console. It seems you added a Cloudflare statistics script that does not work in my browser. This is also a fork of an old version. Where for example "let x=7" in a script
312.
by mg 2y ago
I have various problems with these. That's why I made my own. Some of the problems are: You can't edit the raw HTML because the site injects additional stuff into the HTML you write. Editing is sluggish because of the heavy proces
313.
by mg 2y ago
Another approach is to paste it into an online HTML editor with preview. Here is one that I made: https://no-gravity.github.io/html_editor/ This has the benefit that you can also edit the html afterwards. It also has t
314.
by mg 2y ago
Yes, that might be helpful. First attempt: Insignificant does not mean ineffective Not as good as "correlation isn't causation" though. I wonder if there is a better quip. Or if the concept is too complex to wrap in
315.
by mg 2y ago
There is a lot of confusion on how to interpret studies. The reasoning of this article seems to be "The study did not produce a significant result, therefore the treatment is ineffective". But that is not how to think about signif
316.
by mg 2y ago
I tried my standard testprompt: Two cars have a 100 mile race. Car A drives 10 miles per hour. Car B drives 5 miles per hour, but gets a 10 hour headstart. Who wins? And the Liquid-40B model failed with a long explanation w
317.
by mg 2y ago
Has it been publicly benchmarked yet, if this approach: Hello LLM, please solve this task: <task> Can be improved by performing this afterwards? for iteration in range(10): Hello LLM, please solve this task: &lt
318.
by mg 2y ago
The first example shows a GET request and states that it is "common and interoperable query". But that "if the query parameters extend to several kilobytes or more of data it may not be, because many implementations place lim
319.
by mg 2y ago
TL/DR: Haskell makes you add more meta information to the code, so that compilers can reason about it. One of the examples in the article: Python: def do_something(): result = get_result() if result != "a res
320.
by mg 2y ago
Yes. Say Joe just finished the 1000-add_logout_button.txt task and committed it with the following content: owner: mg info: 2024-09-03 17:02 I added the logout button. Settled on rounded corners which we
321.
by mg 2y ago
Everybody works on a fork on their own machine. Since everybody pulls the changes of the others and sees if something collides with their own changes, its pretty much possible to do it manually. Thats how I started this approach. It's
322.
by mg 2y ago
These days when I collaborate with other developers, we do it all in a git repo that is simply hosted on a VM everyone has ssh access to. Project management is done in a plan/ directory which has task entries like 1000-add_logout
323.
by mg 2y ago
The optimizations the compiler can do because of the "var" the author added in the nim version of the first example should be also possible without it. Because Python defines variables as local or global at compile time. The code
324.
by mg 2y ago
Every now and then I stumble over the problem that in LibreOffice, you cannot date-format the x-axis of a chart. So no x/y chart with lines at the 1st of every month for example. https://x.com/marekgibney/status&#x
325.
by mg 2y ago
He uses "vanilla PHP/HTML/MySQL": https://x.com/AnTheMaker/status/1530851239310802947
326.
by mg 2y ago
Well, you can define the syntax however you like in a new DSL, right? I like the syntax notpushkin proposed and would extend it a bit like this: fruits* = MultiChoiceInput # Pick two out of three fruits Fruits are heal
327.
by mg 2y ago
It looks like it could be made leaner. This: name* = TextInput( | question = What is your name? | description = Let's get started with the survey. First, please tell us your full legal name ac
328.
by mg 2y ago
We research all product data manually and then have AI cross-check the data and see how well it can replicate what the human has researched and whether it can find errors. Actually, building the AI agent for data research takes up most of m
329.
by mg 2y ago
The author gives two reasons why AI won't replace the need for metadata: 1: LLMs "routinely get stuff wrong" 2: "pricy GPU time" 1: I make a lot of tests on how well LLMs get categorization and data extraction right
330.
by mg 2y ago
As far as I know, Alphabet has spent over $10B on Waymo so far. That surely makes it one of the most expensive (in terms of development before reaching profitability) products of all time that were financed by the private sector? I would lo

More ›