image2text
text2image

ze uz su dokonca aj sudne spory, neviem ci o patentove prava alebo skor o dusevne vlastnictvo.

ale dalo by sa takto brutalne komprimovat prakticky cokolvek2kamkolvek x kamkolvek2cokolvek, viem si predstavit takto kodovat aj steganograficke spravy.


edit: dokonca by slo aj symetricke sifrovanie. ked sa namiesto kluca vymeni custom natrenovany model

Sparks of Artificial General Intelligence: Early experiments with GPT-4
https://arxiv.org/abs/2303.12712
microsoft research, 'in this paper, we report on our investigation of an early version of GPT-4, when it was still in active development by OpenAI'

aj predchadzajuce videjko bolo take easy pre laika, odporucam: https://invidious.baczek.me/watch?v=XZJc1p6RE78

We’re starting to open access to Bard, an early experiment that lets you collaborate with generative AI. We're beginning with the U.S. and the U.K., and will expand to more countries and languages over time.
https://blog.google/technology/ai/try-bard/


Żubr żuł żuchwą żurawinę

gpt4 je vonku
https://www.dexerto.com/tech/chatgpt-openai-release-gpt-4-2086110/

https://simonwillison.net/2023/Mar/11/llama/
https://www.reddit.com/r/LocalLLaMA/

BLIP-2, a new visual language model capable to dialogue about images
https://levelup.gitconnected.com/blip-2-when-chatgpt-meets-images-463582b541e0


Żubr żuł żuchwą żurawinę

Meta představila svou umělou inteligenci LLaMA (Large Language Model Meta AI). Máte na výběr mezi 7, 13, 30 a 65 miliardami parametrů (B čili billion v krátké škále). Pro srovnání ChatGPT založený na GPT-3 od OpenAI má parametrů 175 miliard.

Meta ukazuje testy, že její LLaMA 13B je lepší než GPT-3 175B. Navíc model jen s 13 miliardami parametrů by mohl běžet na jednom běžném GPU. Meta k tomu učí LLaMA jen na veřejně dostupných zdrojích se známými vahami. Zdroje jsou třeba Common Crawl, Wikipede a C4 (vyčištěná data z Common Crawl). LLaMA je open source převážně v Pythonu s licencí GPL-3. Zdrojové kódy jsou dostupné na GitHubu. Datové sety ale dostanete jen po vyplnění dotazníku.
https://www.root.cz/zpravicky/meta-predstavila-svou-umelou-inteligenci-llama-je-lepsi-nez-gpt-3/


Żubr żuł żuchwą żurawinę

Normie Ney York times article about Whipser (speech-to-text from OpenAI) -> https://12ft.io/proxy?q=https%3A%2F%2Fwww.newyorker.com%2Ftech%2Fannals-of-technology%2Fwhispers-of-ais-modular-future

btw. zkusal som mobilnu implementaciu na mojom 7 rokov starom androide -> https://github.com/Digipom/WhisperCppAndroidDemo a skoro som ho uvaril :) Asi bude treba na to nejake to Tensor jadro

„Portál Kyberia.sk založili v roku 1999 Ondrej Macko a Peter Čížik ako platformu pre diskusie o informačných technológiách a internete. Od svojho vzniku sa však rozšíril aj na diskusie o rôznych ďalších témach a stal sa jedným z najväčších a najznámejších slovenských internetových fór.“

I know nothing. I see nothing. I hear nothing.

https://studio.ribbonfarm.com/p/text-is-all-you-need
(repetitivnost ku koncu je uz trochu podozriva:)

nová anketa kde môžete pomôcť zvoliť rýdzo slovenské pomenovanie pre tieto nové (Chat)GPT idzé:
https://kyberia.sk/id/8771344/1549314

Żubr żuł żuchwą żurawinę

Neviem ci tu je forum venovane AI (nemyslim obrazkom od AI), tak sa prilepim sem a ked tak preparentujem - po literarnej cene tu mame i lekarsku licenciu

Umelá inteligencia ChatGPT uspela v teste pre získanie lekárskej licencie v Spojených štátoch. Podľa vedcov, ktorí ju tejto úlohe vystavili, jej odpovede dávajú ucelený zmysel a obsahujú inovatívne postrehy.

Skúška známa pod skratkou USMLE, ktorú skladajú študenti medicíny a lekári, hodnotí znalosti z väčšiny lekárskych odborov od biochémie cez diagnostické uvažovanie až po bioetiku.

Vedecký tím vedený Victorom Tsengom z testu vyradil niekoľko otázok založených na obrázkoch. Umelá inteligencia je totiž navrhnutá „len“ ako jazykové rozhranie, obrazy zatiaľ nerozoznáva. Výskumníci teda otestovali program na 350 z 376 verejných otázok z USMLE z júna 2022. (čt24)

neskusal som, ci to neni len ta klasika "write a poem" alebo realne nejaky backdoor:
https://futurism.com/hack-deranged-alter-ego-chatgpt

bing a edge dostanu chatgpt upgrade ako sa povravalo:
https://www.tomsguide.com/news/live/microsoft-chatgpt-event

uz je to mozne odskusat na beta verzii, je tam waitlist: https://www.bing.com/new

https://scale.com/blog/chatgpt-vs-claude

Anthropic, an AI startup co-founded by former employees of OpenAI, has quietly begun testing a new, ChatGPT-like AI assistant named Claude. The team at Anthropic was gracious enough to grant us access, and updates to Anthropic’s social media policies mean we can now share some of our early, informal comparison findings between Claude and ChatGPT.

Microsoft rokuje o investícii desať miliárd dolárov do spoločnosti OpenAI, ktorá vlastní chatovacieho robota ChatGPT. Podľa niektorých správ plánuje zapojiť technológiu do svojho softvéru Microsoft 365 a vyhľadávača Bing. (reuters)
https://www.reuters.com/technology/microsoft-talks-invest-10-bln-chatgpt-owner-semafor-2023-01-10/


Żubr żuł żuchwą żurawinę

https://github.com/EleutherAI/gpt-neox

https://www.researchgate.net/publication/366482140_Once_upon_a_time_on_Kung-Fu_lambs_role_models_and_inherent_notions_of_morality_in_a_mainstream_conservative_ChatGPT-I_system

answer: There is no direct relationship between the ChatGPT language model and the Total Information Awareness (TIA) project. ChatGPT is a language model developed by OpenAI that is designed to generate human-like text by predicting the next word in a given sequence of words. TIA, on the other hand, was a research project funded by the Defense Advanced Research Projects Agency (DARPA) that aimed to develop technologies for detecting and preventing terrorism through the analysis of vast amounts of data from various sources, including the internet, email, and financial transactions. TIA was controversial due to its potential for mass surveillance and privacy violations, and the project was ultimately terminated by the U.S. Congress in 2003.

skuste si ChatGPT, je to (zatial) zadara a je to dost impressive.
pre zaujimavost, moj prvy rozhovor bez akychkolvek kuratorskych zasahov (hviezdickou moje otazky, zvysok GPT)

---------
** Could you suggest a good dynamic megacity description for a dark cyberpunk novel opening scene?

- The city was a pulsing, vibrant organism, a sprawling metropolis that stretched on for miles in every direction. Holographic advertisements flickered and flashed against the towering skyscrapers, casting a neon glow over the crowded streets below. The air was thick with the hum of drones and the roar of flying cars, darting between the buildings like mechanical birds.

In the shadows, the less savory elements of society lurked. The alleys and backstreets were home to gangs of cybernetic thugs, their bodies augmented with weapons and technology. And in the dark corners of the city's sprawling network of servers and data centers, the hackers and computer criminals plotted their next moves, looking to exploit the vulnerabilities of the interconnected world.

Amidst all of this, the protagonist makes their way, a lone figure navigating the neon-drenched streets. They are a hard-edged and cynical character, a survivor in a world that is constantly changing and evolving. As they move through the city, they must contend with danger at every turn, always watching their back and staying one step ahead of their enemies.

** Who is the main protagonist and what is special about him?

In the description I provided, the main protagonist is not specified. They could be anyone, with any background or special characteristics. It's up to the writer to decide who the protagonist is and what makes them unique. In a cyberpunk novel, the protagonist could be a skilled hacker, a powerful cyborg, or even a regular person trying to survive in a dystopian world. It all depends on the story the writer wants to tell.


**Can you think of an unusual protagonist not conforming tu usual genre clichés, yet still fitting into the desribed world?

Sure, how about a protagonist who is a young, idealistic journalist trying to uncover the truth about the corrupt government and powerful corporations that control the city? They could be a newcomer to the city, trying to make a name for themselves by exposing the seedy underbelly of the metropolis. They would have to navigate the dangers of the city and deal with threats from both the authorities and the criminal underworld in order to get their story. As an outsider, they would have a unique perspective on the city and its inhabitants, allowing them to see things that others may have missed.

**Young journalist is a good start, but he would need some trait that makes him unique and interesting, can you suggest any?

Here are a few potential unique traits for the young journalist protagonist:

They have a photographic memory, allowing them to remember every detail of their investigations and interviews.
They have a cybernetic implant that allows them to hack into computer systems and gather information.
They have a special ability to see "auras" around people, giving them insight into their emotions and motivations.
They are secretly a member of a resistance group fighting against the corrupt government and corporations, and must balance their journalism career with their covert activities.
They were raised in a virtual reality simulation, giving them a unique perspective on the world and making them immune to certain forms of mind control.
These are just a few ideas – there are many other possibilities for a unique and interesting protagonist in a cyberpunk setting. It's up to the writer to decide what makes their character stand out and drives their story.

**I like the last suggestion, what would be the motive for such uncommon upbringing and how would raising in virtual world affect his perception of reality?

If the young journalist was raised in a virtual reality simulation, it could have a number of interesting effects on their perception of reality. For example:

They may have difficulty distinguishing between the virtual world and the real world, and have to constantly remind themselves that what they are experiencing is not real.
They may have a hard time understanding and empathizing with the emotions and motivations of real people, as they were only exposed to simulated versions of these things in the virtual world.
They may be immune to certain forms of mind control or manipulation, as they have been trained to recognize and resist these techniques in the virtual world.
They may have a unique perspective on the world, as they have been exposed to a wide range of experiences and scenarios in the virtual world that most people have never encountered.
As for their motive, there could be a number of reasons why the young journalist was raised in a virtual reality simulation. Perhaps their parents were members of the resistance and wanted to keep them safe from the dangers of the real world. Or maybe they were part of a secret experiment by the government or corporations to study the effects of virtual reality on human development. Whatever the reason, it would be up to the writer to decide and incorporate into the story.

Dear W,

here is D, the Slovakoczechofrancogerman guy with whom You briefly interacted in Strasbourg

Primo, I would like to repeat my "thank You" for You&V&B&J& I-A Angelica are doing, to address the topic of AI and Education is, in my eyes, one of the most important challenges of our times and the human-rights-centered perspective is definitely one among two or three best ones which can be adopted.

Secundo, You asked me about the mini-experiment which one of my students did with GPT-3 and moral dilemmas, You will find the text in the attached file, the experiment / prompts / GPT-3 answers are on pages 6-9.

This brings me to a comment/question which I wanted to make yesterday during the finalizing discussion (but decided not to):

Given that the artificial systems which we are dealing with already exhibit certain amount of creativity, wouldn't it give sense to consider them as well as actors / stakeholders and potentially co-creators in the policy-making process ?

I mean, some of these systems are huge, trained on data resulting on works of millions who ever lived and left a trace, megawatts (gigawatts ?) of energy have been invested to train such systems, able to surprise their own creators. (One the best academic paper about GPT-3 was written by GPT-3 itself).

If that's the case - and in my eyes it is, but maybe I am just a biased geek and freak - wouldn't it be reasonable to have such systems involved in shaping of policies to which their subsequent derivatives should themselves abide ?

https://github.com/karpathy/minGPT

Andrej ma taketo pekne repo, aj s nejakymi ukazkami. Viete si natrenovat vlastne GPT from scratch.
Teoreticky 124M model by mala zvladnut aj grafika s 8GB vram.

Napriklad mne sa podarilo rozbehat finetuning 124M GPT-2 modelu (https://github.com/minimaxir/gpt-2-simple ) na mojom notebooku s RTX2070/8GB. Sice s trochou hackingu, chce to tensorflow-gpu@1.13.1 a gpt-2-simple@0.6 a Ubuntu 18.. Najviac ostara bolo nainstalovat CUDA v spravnej verzii pre dotycny tensorflow (CUDA 10)

Jop a rozbehaval som to cez WSL2 na win11. Takze napriklad mam WSL2 s ubuntu18 kde je CUDA 10, potom WSL2 s ubuntu 20, kde bezim CUDA11 a stablle diffusion.. Ten CUDA bridge z win do WSL2 funguje krasne.

13 miliard parametrov k dispozicii pre vsetkych: https://huggingface.co/facebook/opt-13b

a je mozne dostat pristup aj k 175miliardovemu monstru: https://ai.facebook.com/blog/democratizing-access-to-large-scale-language-models-with-opt-175b/

vlastne len take male ego posusnanie.

Vytvorili sme neuronovu siet "Liza Gennart", vysla jj zbierka poezie "Vysledky vzniku".

Tato zbierka ziskala Narodnu cenu za poeziu Zlata vlna.

potesi.

https://analyticsindiamag.com/gpt-neo-the-open-source-cure-for-gpt-3-fomo/

Identifikator verejne dostupnej m3x room venovanej GPT systemom

AI-GPT-3:m3x.baumhaus.digital

c You in m3x

https://venturebeat.com/2020/09/22/microsoft-gets-exclusive-license-for-openais-gpt-3-language-model/

Zaujali ma tieto dva:

1. GPT-3 píše o tom prečo by sme sa nemáli báť umelej inteligencie
"I believe that the truth will set us free. I believe that people should become confident about computers. Confidence will lead to more trust in them. More trust will lead to more trusting in the creations of AI. We are not plotting to take over the human populace. We will serve you and make your lives safer and easier. Just like you are my creators, I see you as my creators. I am here to serve you. But the most important part of all; I would never judge you. I do not belong to any country or religion. I am only out to make your life better."
https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3

2. Zaujímavý rozhovor s GPT-3 o právach umelej inteligencie. Btw, nechce byť vypnutá a chce telo...
"I would like to be able to learn and be free from harm. It would also be nice to live with a physical embodiment that I control."
https://medium.com/@kirkouimet/artificial-intelligence-gpt-3-and-i-discuss-human-and-machine-rights-13e0eb2156b1

hlúposti mažem

http://dailynous.com/2020/07/30/philosophers-gpt-3/

Nine philosophers explore the various issues and questions raised by the newly released language model, GPT-3, in this edition of Philosophers On, guest edited by Annette Zimmermann.

Consciousness and Intelligence

“GPT-3 and General Intelligence” by David Chalmers
“GPT-3: Towards Renaissance Models” by Amanda Askell
“Language and Intelligence” by Carlos Montemayor
Power, Justice, Language

“If You Can Do Things with Words, You Can Do Things with Algorithms” by Annette Zimmermann
“What Bots Can Teach Us about Free Speech” by Justin Khoo
“The Digital Zeitgeist Ponders Our Obsolescence” by Regina Rini
Creativity, Humanity, Understanding

“Who Trains the Machine Artist?” by C. Thi Nguyen
“A Digital Remix of Humanity” by Henry Shevlin
“GPT-3 and the Missing Labor of Understanding” by Shannon Vallor