dsilverz

joined 2 months ago
[–] dsilverz@thelemmy.club 24 points 10 hours ago

Have you ever heard of the Riemann hypothesis? Since 1859 it's yet to be solved. The generalization of prime numbers (i.e. a function f(n) that yields the nth prime) would impact fields such as Navigation Systems and Traffic Management, Communication Systems and Satellite Communication (i.e. your Internet connection could become more efficient and faster), Astrophysics and Cosmology, Quantum Mechanics, AI and Machine Learning, E-commerce, Finances and Algorithmic Trading, among many other fields. (Yeah, it seems like nothing. /s)

[–] dsilverz@thelemmy.club 14 points 18 hours ago (1 children)

Just a tip for the developer/sysadmins of loops.video, as a developer myself: Seems like loops.video has no DKIM or SPF configured (if that's the domain being used to sent activation links/codes), so the tendency is for most email providers to block the mail or move it straight to Spam folder. The situation worsens when many users try to sign-up for an account, so loops.video sends a lot of sequential emails (which is something that could be seen as "spam behavior" by email providers). The developer should ensure that mail delivery is properly configured, particularly the trust headers (DKIM and SPF, as mentioned before) needed for sending emails.

[–] dsilverz@thelemmy.club 6 points 18 hours ago

Seems like loops.video has no DKIM or SPF configured (if that's the domain being used to sent activation links/codes), so the tendency is for most email providers to block the mail or move it straight to Spam folder. The situation worsens when many users try to sign-up for an account, so loops.video sends a lot of sequential emails (which is something that could be seen as "spam behavior" by email providers). The developer should ensure that mail delivery is properly configured, particularly the trust headers (DKIM and SPF, as mentioned before) needed for sending emails.

[–] dsilverz@thelemmy.club 10 points 18 hours ago (5 children)

Just a tip: always strip the si=... parameter from URLs. The si parameter is a form of tracking from platforms (particularly Youtube and Soundcloud, although other platforms use it as well), so the platforms can know who shared things and who opened the things that the former shared. When you send a clean link, without the si parameter, it's more difficult (if not impossible) for platforms to determine who shared the link being opened/downloaded.

[–] dsilverz@thelemmy.club 5 points 1 day ago

From knives to ammunition and missiles, all these things are made with rocks (minerals) so, in a sense, humans still use rocks to fight each other. As they say: "War... War never changes..."

[–] dsilverz@thelemmy.club 66 points 1 day ago (2 children)

I wonder why disinformation and misinformation is such a problem nowadays... Maybe the access to scientific papers should be opened and democratized so everyone, regardless of social and economic classes, could read and lookup reliable knowledge? Nah, just paywall 'em all and blame those silly conspiracy theorists for online misinformation, it'll certainly work. /s

[–] dsilverz@thelemmy.club 5 points 1 day ago

I'm not sure if it's a São Paulo (as in the state, not the city) thing, but I had English classes when I was in public high school ("ensino médio"). They weren't the best English courses out there (i.e. they weren't comparable to Brazilian schools that specialize in English courses such as CCAA, CNA, Fisk and Wizard), but they offered a good start for those who had no prior knowledge of the English language. It's also worth mentioning that people who work in IT have more potential to come into contact with communication in English because a lot of documentation is in English. But I totally agree with you that most of the population does not have quality access to English courses.

[–] dsilverz@thelemmy.club 10 points 1 day ago

I'm Brazilian. For personal projects and snippets, especially if I'm going to share their code publicly (e.g. GitHub or GitHub Gists), I often use English. However, when it's a project from a company I'm working for, I use Portuguese, as every company I've worked for so far are Brazilian (and my coworkers were Brazilian as well).

[–] dsilverz@thelemmy.club 7 points 1 day ago (1 children)

Just a small Portuguese correction: "Bem vendos aos fediverses" should be "Bem-vindos aos fediversos!".

[–] dsilverz@thelemmy.club 6 points 1 day ago (1 children)

Despite the lack of apps, Windows Phone was very good for me at that time, as I had two Lumias. They were quite cheap but rather powerful (again, despite the lack of apps like internet banking, but they did have Whatsapp and Telegram). I left WP and Lumia when Whatsapp ended its support for WP in December 2019 (if I remember correctly), and Nokia's Android phones were expensive at the time, so I tried the Asus Zenfone (because I see Asus as a good PC hardware manufacturer). Two years later, my Zenfone started to drain faster because the battery started to swell, so I bought a Nokia with Android, which I still use nowadays. This latest acquisition made me realize that, indeed, Nokia is no longer the same: although it has the Nokia's bold design ("almost indestructible"), it is a slow smartphone. I fixed my Zenfone battery and used both phones simultaneously for another two years, when the Zenfone battery stopped holding a charge again (although, this time, it didn't swell). Since I couldn't find a replacement battery for the Zenfone, I stuck with the Nokia, but soon I'll try another brand like Xiaomi, or maybe Asus again since my previous experience with a Zenfone was really good.

[–] dsilverz@thelemmy.club 4 points 2 days ago

Some examples that I remember are:

[–] dsilverz@thelemmy.club 2 points 2 days ago

Some surrealist (not exactly "gibberish" in the literal sense) ideas:

  • "Let ᚠ be the ζth factor of the ξth Pontryagin dual element from a Laplacian matrix, hence, the numerical representation from a graph, a Pontryagin duality graph. Let Σᛇ be the sum of probabilities such as ᚠ equals to zero. Determine the probabilities for ᛗ considering that sinh(ᛗ-ᚠ) × ᛟφv² + 1/log(dx) = φͲδx³ + ᚠδx² + 2x where δ is the Gompertz constant and x is the nodal variation for each parallelogram axe."
  • "Given that a conventional passenger airliner flies at speeds below Mach 1, what appears to have been the exact sequence of events that led to an Airbus A380 stalling on August 23, 2027, when a flight (whose flight recorder was recovered but was severely damaged internally) carrying 138 passengers crashed into the Indian Ocean during a strong CME that somehow caused the plane to exceed Mach 1 before its crash?"
  • "Derek is wandering at the cemetery during midnight. He ate cooked rice and oat flour in the previous day. His cat, Mower, was diagnosed with pancreatitis. The entire Northern Hemisphere is announced to face severe weather due to anomalies within the Gulf Stream. Back at the cemetery, a specific grave seems misplaced: the gravedigger dug through a water pipe, now the grave is overflowing and filled with dirty water. Why those ravens seem to be following Derek?"
 

cross-posted from: https://thelemmy.club/post/17993801

First of all, let me explain what "hapax legomena" is: it refers to words (and, by extension, concepts) that occurred just once throughout an entire corpus of text. An example is the word "hebenon", occurring just once within Shakespeare's Hamlet. Therefore, "hebenon" is a hapax legomenon. The "hapax legomenon" concept itself is a kind of hapax legomenon, IMO.

According to Wikipedia, hapax legomena are generally discarded from NLP as they hold "little value for computational techniques". By extension, the same applies to LLMs, I guess.

While "hapax legomena" originally refers to words/tokens, I'm extending it to entire concepts, described by these extremely unknown words.

I am a curious mind, actively seeking knowledge, and I'm constantly trying to learn a myriad of "random" topics across the many fields of human knowledge, especially rare/unknown concepts (that's how I learnt about "hapax legomena", for example). I use three LLMs on a daily basis (GPT-3, LLama and Gemini), expecting to get to know about words, historical/mythological figures and concepts unknown to me, lost in the vastness of human knowledge, but I now know, according to Wikipedia, that general LLMs won't point me anything "obscure" enough.

This leads me to wonder: are there LLMs and/or NLP models/datasets that do not discard hapax? Are there LLMs that favor less frequent data over more frequent data?

 

First of all, let me explain what "hapax legomena" is: it refers to words (and, by extension, concepts) that occurred just once throughout an entire corpus of text. An example is the word "hebenon", occurring just once within Shakespeare's Hamlet. Therefore, "hebenon" is a hapax legomenon. The "hapax legomenon" concept itself is a kind of hapax legomenon, IMO.

According to Wikipedia, hapax legomena are generally discarded from NLP as they hold "little value for computational techniques". By extension, the same applies to LLMs, I guess.

While "hapax legomena" originally refers to words/tokens, I'm extending it to entire concepts, described by these extremely unknown words.

I am a curious mind, actively seeking knowledge, and I'm constantly trying to learn a myriad of "random" topics across the many fields of human knowledge, especially rare/unknown concepts (that's how I learnt about "hapax legomena", for example). I use three LLMs on a daily basis (GPT-3, LLama and Gemini), expecting to get to know about words, historical/mythological figures and concepts unknown to me, lost in the vastness of human knowledge, but I now know, according to Wikipedia, that general LLMs won't point me anything "obscure" enough.

This leads me to wonder: are there LLMs and/or NLP models/datasets that do not discard hapax? Are there LLMs that favor less frequent data over more frequent data?

 

Firstly, sorry if this is not the adequate place for my question; if it's the case, let me know.

The title may seem confusing, so let me detail it: I'm more of a commenter person, and some of my comments are replied, and Lemmy notifies me of those direct replies. However, there are moments when those replies receive third-party replies, so my comment turns into some kind of "sub-thread", something that's interesting for me to read and follow. For those third-party replies, I don't receive notifications, so I have to access each direct reply that was notified so to find possible "sub-threads".

There seems to me to be no option to "receive notifications for this post/comment/reply", only the automatic opt-in of notifications for direct replies.

So really isn't there such an option? Or is this an instance-specific feature and the instance I belong to (thelemmy.club) don't have it?

view more: next ›