Agraillo

14 ms·

121.
by Agraillo 3y ago
XPath is very impressive and expressive when one just looks outside the xml/html world. An abstract XPath query selector only needs a tree enumerator with nodes reporting the name and (optionally) some attributes. One of such applicati
122.
by Agraillo 3y ago
Very impressive, actually it's a step in a direction of visualizing music, i.e visual means to create a repetitive patterns in the brain similar to those created when we listen to musical pieces. Some constraints (like measures) or qua
123.
by Agraillo 3y ago
I don't think 11yr vs 15yr/16yr is a big difference for measuring. I started early, probably falling into 11yr category. Many years passed, and recently I had a chance to look at my code of the age probably around 20. There were s
124.
by Agraillo 3y ago
I'm sure the LLM-based engines will shine here, partly they are already here. A couple observations: - Google Lens, now by default activated when you post an image to Google Images ( https://images.google.com ) has a text rec
125.
Baby Tooth Survey(en.wikipedia.org)
3 points by Agraillo 4y ago|0 comments
126.
by Agraillo 4y ago
Recently I started to think that somehow humans tend not to rationally accept or comprehend ideas but to like, love and worship them. I'm not an exception either. The other day after reading about Tensegrity[0], I managed to make a sim
127.
by Agraillo 4y ago
Thanks for the article, the ending is interesting because it mixed "Not Even Wrong" and Wronger than Wrong" (still wrapping my head around it): "But if you believe that thinking science is unbiased is just as wrong as th
128.
by Agraillo 4y ago
> is that conversations about what is wrong and how wrong it is depend so heavily on the very first thing you put your faith in Rationalwiki article [0] about the topic has a nice epigraph ”It’s a little wrong to say a tomato is a vegeta
129.
Wronger Than Wrong(en.wikipedia.org)
137 points by Agraillo 4y ago|86 comments
130.
by Agraillo 4y ago
The "1 in 5" example is the most prominent one (introduced at the start of the article), but I was more impressed with another one - "Do you say the word hello or charisma more". Similar is this. Try remembering two of y
131.
Frequency Format Hypothesis(en.wikipedia.org)
47 points by Agraillo 4y ago|11 comments
132.
by Agraillo 4y ago
Not specifically LLM, but it looks like generating fonts with GAN is the subject ot this work: https://arxiv.org/abs/2203.10348 "Our goal is to generate fonts with specific impressions, by training a generative ad
133.
by Agraillo 4y ago
Thanks for the video. What is also interesting is the episode prior to 2:19:... when two month before the war Bennett met Putin in Sochi. Zelensky wanted to contact Putin and meet him possibly to prevent the upcoming events. But to do this
134.
by Agraillo 4y ago
> Imagine what it would look like if ChatGPT were a lossless algorithm. If that were the case, it would always answer questions by providing a verbatim quote from a relevant Web page. We would probably regard the software as only a sligh
135.
by Agraillo 4y ago
At least for this audience there's much less sophisticated algorithm for discovering - F5 on Hacker News
136.
by Agraillo 4y ago
Your comment inspired... HN is a pool of ideas of "above average" quality fighting for the attention of readers in evolutionary sense. As in the life they born and die, the lifespan is very short, but the "DNA" of the be
137.
by Agraillo 4y ago
Good point, I thought that this design belongs to the print media world, when the media is split by pages, not to the scrolling continuous world of the web.
138.
by Agraillo 4y ago
Reading, watching the video, I keep wondering whether the authors just rehashed the ideas from their native country's bestseller - QualityLand by Marc-Uwe Kling
139.
by Agraillo 4y ago
A random jumper of a good curated list of interesting/funny articles on Wikipedia, but.. But my guess (based on reverse Birthday Problem) is that the list is too short, about 300-500 links (since I met the same article after about 20-3
140.
by Agraillo 4y ago
I suspect with AI we always will be in "this story (song, picture) is Ok" and never "is great" category. I recall the member of ABBA was asked whether they ever felt the song would be a hit. She (this was definitely eith
141.
by Agraillo 4y ago
> The original version only supports *.wav audio files with 16kHz sample rate This particular point is not true (at least not fully). The version publicly announced in 2022 had ffmpeg dependency for supporting any audio-containing format
142.
by Agraillo 4y ago
So many links... great resource. But I'm going to be a little off-topic. This is a second time today when the page posted on HN contains a reference (link) to HN itself, a recursive binding in a sense. It's an interesting phenomen
143.
by Agraillo 4y ago
Good point. That's probably why high-quality tv series emerged as a phenomenon. Once the TV producers tried making expensive, star-fronted, good-story series, they've got a new hit-driven ecosystem.
144.
by Agraillo 4y ago
Idea came to be used as a captcha, though the list of choices should be extended from (cat, dog, bird) to restrict bots random guessing. The task is not that easy, but way more entertaining than finding bridges and pandas or entering nonsen
145.
by Agraillo 4y ago
An old joke totally applicable to Mr. Wolfram: "Such a shame we're talking about myself, let's talk about you. Have you read my recent book?"
146.
by Agraillo 4y ago
Just to save the time for others, NIOSH Sound Level Meter is iOS only
147.
by Agraillo 4y ago
Good point and that's why I recently came to an interpretation that search engines return some kind of "tables" for us (but not in the tabular form). List of links are rows, and for every link we have properties ("column
148.
by Agraillo 4y ago
Thanks for mentioning. Interestingly, I recently tested several systems with identical questions ( https://www.perplexity.ai/ , https://www.phind.com (formerly beta.sayhello.so). The Kagi beta is sometimes on par,