Daemon Silverstein

“Press F to pay respects” Same vibe

I’m a 10+ (cumulative) yr. experience dev. While I never used The GitHub Copilot specifically, I’ve been using LLMs (as well as AI image generators) on a daily basis, mostly for non-dev things, such as analyzing my human-written poetry in order to get insights for my own writing. And I already did the same for codes I wrote, asking for LLMs to “Analyze and comment” my code, for the sake of insights. There were moments when I asked it for code snippets, and almost every code snippet it generated was indeed working or just needing few fixes.

They’ve been becoming good at this, but not enough to really replace my own coding and analysis. Instead, they’re becoming really better for poetry (maybe because their training data is mostly books and poetry works) and sentiment analysis. I use many LLMs simultaneously in order to compare them:

• Free version of Google Gemini is becoming lazy (short answers, superficial analysis, problems with keeping context, drafts aren’t so diverse as they were before, among other problems)
• free version of ChatGPT is a bit better (can keep contexts, can issue detailed answers) but not enough (it does hallucinate sometimes: good for surrealist poetry but bad for code and other technical matters when precision and coherence matters)
• Claude is laughable hypersensitive and self-censoring to certain words independently of contexts (got a code or text that remotely mentions the word “explode” as in PHP’s explode function? “Sorry, can’t comment on texts alluding to dangerous practices such as involving explosives”, I mean, WHAT?!?!)
• Bing Copilot got web searching, but it has a context limit of 5 messages, so, only usable for quick and short things.
• Same about Bing Copilot goes for Perplexity
• Mixtral is very hallucination-prone (i.e. does not properly cohere)
• LLama has been the best of all (via DDG’s “AI Chat” feature), although it sometimes glitches (i.e. starts to output repeated strings ad æternum)

As you see, I tried almost all of them. In summary, while it’s good to have such tools, they should never replace human intelligence… Or, at least, they shouldn’t…

Problem is, dev companies generally focus on “efficiency” over “efficacy”, wishing the shortest deadlines while wishing some perfection. Very understandable demands, but humans are humans, not robots. We need our time to deliver, we need to cautiously walk through all the steps needed to finally deploy something (especially big things), or it’ll become XGH programming (Extreme Go Horse). And machines can’t do that so perfectly, yet. For now, LLM for development is XGH: really fast, but far from coherent about the big picture (be it a platform, a module, a website, etc).

While it offers a concurrent alternative to Google translate, it still lacks some features, as @[email protected] mentioned, many languages are missing. In my case, I sometimes experiment with terms across various languages, sometimes Hindi (“O param Devi Kaali”), sometimes latin (“Vita mortem manducat, Mors manducat vitam” is a latin phrase I wrote myself, following Latin grammar rules), sometimes Hebrew (especially for Gematria calculation using numerical values from Hebrew letters (Aleph is 1, Bet is 2, Gimmel is 3, and so on) after translating/transliterating a word/name such as “לילית”). For these kinds of experimentation, DeepL can’t really be of use, so I need either Google Translate or Bing Translate (both support the aforementioned languages).

Exactly. Also, it doesn’t have Latin (used for both scientific terms such as “Athene Cunicularia”, philosophical such as “Homo homini lupus est”, as well for liturgical and ritualistic texts, especially occult texts) nor Hebrew.

Google translate does this, too. This not only pollutes the back button, it also pollutes the entire browser history.

Randomness (even the computer’s pseudorandom) is really amazing. Perlin noise, Sierpinski triangles traced through random walking, etc… Lots of things can be done with random sequences.

Yep. The lemon-flavored soda. Its fizzling sound.

Also, Odyssee (at least Louis Rossmann, I follow him through Odyssee)

Arch Linux on a daily basis of the night sky is upwards of the night sky is… (then Gboard enters in a loop about “the night sky is upwards”)

Edit: With “is” Arch Linux is configured to use it for a while the interior of the night sky is upwards of the night…

I want it for KDE lol

Actually, there is a compilation of all the standards specifications. It’s on W3 (World Wide Web Consortium), where all the technical details are deeply documented (called “Technical Reports”), available on https://www.w3.org/TR/ . To this day, there are 309 published Technical Reports regarding “Standard” specifications.

Fun fact: while seeking for the link to send here, I came across a Candidate Standard entitled “Web Neural Network API”, published exactly yesterday. Seems like they’re intending to implement browser-native neural network capabilities inside Web specifications, and seems like the “closer future” I mentioned is even closer… 🤔

It became difficult as Web technologies grown complexier, such as implementing native CPU instructions through WASM, bluetooth through Web Bluetooth, 3D graphics through WebGL, NFC, motion sensors, serial ports, and so on. Nowadays, it’s simply too hard to maintain a browser engine, because many of the former alternatives were abandoned and became deprecated.

remove the limits for ‘trusted’ advertisers

Exactly… Including themselves, as they are a major player in advertising market (Google Adsense).

Not yet, but I often code myself some experiments involving datasets (i like to experiment with Natural Language Processing, randomness, programmatic art and demoscenes, the list goes on).

Dev here. Javascript engines (especially Chromium) have a memory limit (as per performance.memory.jsHeapSizeLimit), in best case scenarios, 4GB max. LocalStorage and SessionStorage (JS features that would be used to store the neural network weights and training data) have even lower limits. While I fear that locally AI-driven advertisement could happen in a closer future, it’s not currently technically feasible in current Chromium (Chrome, Vivaldi, Edge, Opera, etc) and Gecko (Firefox) implementations.

A mix between the last one and the previous. The key error should be set in order to indicate the presence of an error, while status and code represents, respectively, the error numerical code and the error type, while the error message is supposed to be concatenated to an human-friendly error text.

What if you close the app through drawer, and then relaunch it?

Let me try with the three slots from my Gboard’s auto completion:

Woman are the same author of this text out to be the initial words that rhyme fill it up on my own machine learning trying not a spiritual practitioner jobs o seguinte sonho e a minha visão… (then Portuguese words start to be suggested at the first slot from Gboard’s auto complete)

Woman are you doing obsidian mirrors started to get to the nothingness and the moon is more than I do in the morning with the nature of the night sky is upwards of the night sky is upwards… (then it gets stuck in an infinite loop at the second slot from auto complete)

Woman are a form for a wizard 🪄 (and it ends at the emoji at the third slot from auto complete)

Now for the plural women:

Women in a bit I can feel it was just my own personal email to you can notice that I can’t tell if a letter from a random list for example when you have time and the other two images are AI-generated summary whenever I can feel it is the same author analyze it is a text to retrieve it is the same author of this specific picture just a bunch o seguinte sonho… (same situation from first slot with Portuguese words starting to get suggested)

Women that are upwards of the night sky is upwards of the night sky is upwards of the night sky… (same situation from the second slot regarding the infinite loop at the “night sky is upwards of the” sequence)

Women and children of light 🕯️ (Third slot, also ends in emoji)

(For context: I’m sometimes a poetry writer, and I use a lot of LLM to analyze the texts I made, as well as generative AI for generating goth art! also I’m pagan and occult practitioner).

I’m Brazilian so I generally use public Brazilian (Portuguese) groups, publicly available through Telegram searches such as “Livros grátis” (“Free books” in Portuguese) or “livros pdf” (“pdf books”). There probably are similar anglophone groups as well.

Firstly, I search for the book among Telegram book groups. Then, Google/DDG/Bing/Yandex results for “[title of the book i’m searching] filetype:pdf”. If neither yields the book, I turn on my Tor and try to find it inside Imperial Library of Trantor. Worst case scenario, I try to find a working Z-Library HS. But some comments here added to my possibilities, such as Libgen (I knew Libgen from onion, but it’s interesting to see it also routeable on surface).