“Llama 3 with Agents gives you Godlike power” - Pietro Schirano
Вставка
- Опубліковано 23 кві 2024
- LAST 5 DAYS! Get a custom AI plan tailor-made just for you by joining my community: www.skool.com/new-society
Follow Pietro: / skirano
Try EverArt yourself: www.everart.ai/
Maestro Agent framework: github.com/Doriandarko/maestro
DesignerGPT: chat.openai.com/g/g-2Eo3NxuS7...
Follow me on Twitter - x.com/DavidOndrej1
Please Subscribe.
This is an interview with Pietro Schirano, the founder of EverArt.
David Ondrej Podcast - Episode 09
🔥 LAST 5 DAYS! Get a custom AI plan tailor-made just for you by joining my community: www.skool.com/new-society (price increasing soon)
Wow... an amazing and deep conversation. Some of which really opened my eyes to the concepts and history behind the evolution of AI. The passion expressed on the topic and for entrepreneurship in general - literally reignited my own desire to create something new and unique. (FYI - I'm a 66 yr old with an Electronics/Computer background. Military trained or self-taught. AI is not an 'age thing' if you keep your mind open and embrace knowledge. The near future is going to be amazing).
This channel is incredibly impressive! I thoroughly enjoyed this episode. Pietro Schirano is such a kind, thoughtful, and helpful guest; truly a hell of a man! The insights and knowledge shared in this discussion were seriously top notch. Thank you for bringing this to us, brother!
Also, nice to meet you, Pietro!
I live in Dallas, TX, and work as a civil engineer and landscape architect. I’m developing AI products for our industry, which is nearly entirely untapped, especially by insiders like myself. There are so many opportunities because my field is at the forefront of ecological sustainability, which is becoming extremely important.
We live in such an indoor society that people don’t often think about or care for nature, but once you realize the power of nature, you’ll learn to respect it. Man made creations can be destroyed in an instant, nature, she don’t play! lol
My point is, there are countless opportunities to develop tools, applications, and AI agents for my line of work. I’ve been learning to develop for years now. At university, we had to learn Visual Basic and C++, which was from 2007-2012, and now it has come a long way. I’m teaching myself to code well with Python and LLMs! I’m a major supporter of the open-source community.
Thank you, brothers!
This is the best of you podcast, David. The most I loved was improvisation, sincerity and authenticity of both of you. And thanks to you I discovered @Pietro, such a kind and truly inteligent person. Thank you! 😊
Thanks for having Pietro! I was already a fan just reading his tweets, and now I am big time on team Pietro and more and more on team David too!
I'm surprised no one talks about this but as he puts it open AI lobotomised gpt 4 and we had confirmation for it. Before they released gpt 4 they had researchers testing it and it was 30% better than the gpt4 turbo available today. If they release gpt4 at full power today they could call it gpt 4.5. THEY HAD THIS MODEL FOR OVER A YEAR! This we know for sure, don't know anything about the other models they have but they're years ahead of the curve. They just release the minimum possible to not scare people. That's why when other companies announce something open ai announce something better, it's because they are sitting on years of advancements
Yeah, sure 😂😂😂😂
What a wonderful and thoughtful interview. I have been following Pietro for a while, his passion for the space really shines through
Thank you for this. I have been feeling demotivated by my surrounding lately, but you made me realize something about myself again. I was a passionate person before wanting to do something for others using my mind. But because of the stigma here on being smart is being shamed at, i decide to hide this instead. But I know I can do so much given the right conditions and position. Thus, I will try again.. this time i will push myself to try, and learn not for others but for myself
Solid interview! Worth the sub 🙏🏽💎
This as Podcast, AUDIO only! -> GOLD!
When you start your Podcast?
I agree with the skill learning now thinking toward what’s coming. I”m a lot older than you so had my first iPhone when I was 40. I loved hearing how this sparked your future. I am into AI but not as a programmer - as a tool. I minored in computer science in the early 80s learning with punched cards and the later, the Apple II. Imagine how frustrating - redoing cards to fix a single syntax error and figuring out where in the stack it is. Also lived in Italy as a kid. Languages, exposure affects thinking. Point being, understanding how earlier computers operated and early languages (e.g., Fortran) helps understand how to use AI and how to think about what is coming next.
Cool. More Pietro Schirano
Yall aint ready for gpt 5 in a few months...claude wont be even close
I hope sooner than later.
Doubt it
You know how?
And then Claude 4 comes
That's true but Gemini 1.5 ultra will be close to it, and Gemini 2.0 pro may beat it. 😂😂
Open AI is alrey very slow.
Google is ready to fight even Sora now 😂
I switched from GPT4 to Mistral Large - it's definitely better. Comparing it to LLama3 now, we will see
It really strikes me every time I see a video of yours how confident you talk about models and their lack of consciousness because "they are just an LLM." When more senior people in the AI community talk about this issue with much more nuance and care given that we don't quite understand these models as much as we would like, I wouldn't be so categorical and dismissive about LLMs' ability to develop consciousness or for some models to possess some degree of it. At the very least, I would leave it as an open question, especially when science doesn't even have a very clear definition of consciousness to begin with.
Every being has soul which is reflecting through mind and body. Consciousness is achieved through this process. In my opinion models won’t achieve this level of consciousness any time soon all those models do are creating function based on parameters. Back propagation and other algorithms may give an illusion that model has conscious sometimes but it’s just moving function based on data
Today's science is pretty dumb so I wouldn't look there but we know what consciousness is and it is inherent to life itself, don't worry about artificial consciousness, it won't happen.
Lol you guys dont understand, when they will slove memory issues you will see how the robot will be conscious. This is the only problème right now if they can fix this for the robot to remember everything we done. And i think we can alredy do this they just put a time gate on all of this.
Thank you!
Great discussion guys! 👍
Pietro, on your Everart website you need to do more development with "Train AI on any product, style, or mood board. Create content like you've never seen before." I went to your website while listening to the video and didn't understand what it was about until later when you explained it at the end of the video. Or, maybe you can create some better image examples. 🙂
Amazing repos Pietro! Thanks! 😁
David, great job, however, when you were first typing your notes it looked like you were having a chat with someone and not paying attention to what Pietro was saying. It was kind of distracting. In the future, perhaps you could announce that you'll be typing your ideas out as the interview proceeds. That way everyone will know what you're doing and simply think about what a great multitasker you are instead of being distracted! 😁
Yes! Definitely do some more of these! You both feed off each other very well and it makes for very stimulating entertainment along with being highly educational! 😁😁
Thanks for uploading
12:07
Semicolons....
Great interview gents
Lovely!
Thanks a lot
Nutrition for thought! Thank you for this
Interesting discussion on fine-tuning. I see it kind of like how an old HDD file system needs an occasional defrag to organize. The fine-tuning process also mirrors how we need sleep and dreams to sort and incorporate new information in the brain. High context models with RAG capabilities are similar to how we are when we are in the initial stages of learning new info when and can't yet converse intelligently and concisely about a newly learned topic, however after lots of time, thought, and discussion we grok that ability. In the future I suspect that AI will self-monitor it's internal processes and retrain/fine-tune themselves automatically from time to time based on amount of new data that they deem important to incorporate into their evolution.
5:54 The middle model is absolutely critical for affordability and privacy, and speed tradeoffs. Not everybody can afford bigger memory GPUs to fit bigger models, so more granularity lets users use whatever card they have without having to use cloud services.
Some also want a model that is a bit smarter than 8B, but is still faster than 70B.
So there are a LOT of people who are waiting for 13-14B and 30-40B.
I can run Llama-3 70B on my laptop. It spits out it's response slowly; but, it works very well - and cost me nothing.
Your laptop specs?
@@avinashkumarmandavilli8623 I9 CPU, RTX 4070 (8Gb VRAM) and 64Gb System RAM
thanks!
Doesn’t Llama 3 have a low context window?
Transcripts!!! 🎉
Have you tried Phi3? seems pretty good
My guy looking like a young Trent Reznor in this 🤰🏻…
hi could talk about fet,,ocean agix ,all are merging into 1 ai token( I believe)?
Oh no, someone has been in a punch on!
I have worked for large companies, but I love working with small companies. I want to bring AI to them. I will sign up for your course if you can help me keep the internal data secured while accessing these different AI models and frameworks. Is that possible? I'm a software architect, slow on the draw with AI. Your vids are very encouraging to me. Thanks.
A thought just hit me, maybe I can create an obfuscator for personal info for use with commercial AI. Just a thought.
Gemini Ultra is still the best at writing fiction, as far as I can tell.
can running locally?
broo im too broke to get into new society lemme slide in u gots plenty ppl now
i swear i dont know shiit but i will join
What is the thing on your nose? I'm really curious 🤔
it is a device to help breathing
All the cool AI people are doing it. Claude said it makes you smarter
For the first 5 minutes I was thinking wow what is this cloud llm that I have not tried... :D
Me too. Claude daaah
I need someone whos familiar with making agents to check something I've been working on
Would be good to see something actually useful built with with these tools. People always demonstrate building pythons snake game or something like that, which has no real life use.
None of them will be able to achieve AGI first.
Not even Tesla or Grok.
It will be a start-up.
Mark my words. 😊😊
We shall see
@David, a suggestion for another podcast: bring someone who is a sustainability or climate change or circular economy expert and knows AI to discuss how AI, with specific tools and examples, could help save the World. Afterall, we all, even with the bright future of emerging tech, will still be depending on livability conditions on the Earth. Thanks 🫱
Hello, I would like to ask about the community on Discord.
I want to develop a voice assistant for commercial purposes that will work for various enterprises that need call centers.
Can your community help me in this situation?
Bad llama!
The emperors new groove...
I had a flashback
linguaggio di macchina ahahhaha I can totally relate LOL
the true is agents came before llms
Interesting interview but try and look at the screen and be still while listening to a guest otherwise you don't look engaged at all.
He is taking notes. I bet he could get an llm to do that
Moloch
First 🏁
Second
Don't use God's name in vain
What is God's name?
@@honkytonk4465Isis is one of them. But in the context of AI I’d probably go with Sofia - Greek god of knowledge
how does this channel give such scam vibes
going full satanic now? depending on the next video I'm unsubscribing.