Let's not forget the amazing thing that is Control Networks. This thing has made Stable Diffusion that works in the Automatic1111 webui (maybe in others too) far surpass everything else in image generation, it's not even a contest. Also, uncensored models will be the winners in the end regardless of performance (which also helps with that). Until the big companies understand this, they have no chance. I much prefer using the 7b Pygmalion model or the 13b Wizard on my pc (even better with TavernAI for RP or story narration), than a restricted online model that half of the time will output that as an AI model it can't answer... Plus, I get the benefit of privacy with them. edit: The best thing will be a 13-20b well-trained uncensored model with 8-12k tokens and self-reflection that can run on a 10gb GPU, and can also use other tools like calculators, code editors, etc.
"The future of a war is a war that takes a millisecond, the whole thing occurred in a millisecond. That's faster than human decision making time, which means that our defensive systems are going to have to be on a hair trigger. And they're going to have to be invoked by AI, that we don't fully understand." ~Eric Schmidt, Former Alphabet CEO, 03/23/2023
Yes, it is kind of scary that we have to give up some control to these ai models since they need to be able to react extremely quickly. The arms race never stops
I retired in 2020 and I've been away from tech since then. I must confess that I understood almost none of this video. The only takeaway is that AI has become easy for anyone to develop, and Google is fearful of losing its edge ... because others can now offer without restrictions what Google would want to offer with restrictions. Aside from that I don't know wtf this is all about.
You are awesome bro....I am in uni now and I have been learning more about AI lately.I must admit that I never knew that this kind of progress has been made in the field within such a short time.Thanks again for the insights
After the excitement of early 2000s about a free and prosperous future where open source and all (free unregulated internet, no IP, no censorship, the low overhead manifesto and criptopunk), the internet was apparently tamed by corps and the law in about 2008 onwards. Now there is a revival and it AI based one at that! What a time to be alive!
This proves we can do everything, if we work together in a free way and that everyone can be an important help, if we are open to each other. While we are more than just spoken words in animated videos or data. The magic that happened when people really come together, not online, I mean in real life, is something that can not be programmed as we not even close to know what the magic causes, it's just there, because we are more than just a thinking mind.
It does appear that the big companies will be able to use regulations in order to build a moat so that startups and small companies won’t be able to compete without huge amounts of funding
I have no idea why you were recommended, and usually I hate these kinds of commentary videos, but it was nice being able to do something else while you read through it. Thanks! :)
Important note: It is not "Google" and maybe only 1 Google Researcher who thinks this... As stated on SemiAnalysis "The text below is a very recent leaked document, which was shared by an anonymous individual on a public Discord server who has granted permission for its republication. It originates from a researcher within Google. We have verified its authenticity. The only modifications are formatting and removing links to internal web pages. The document is only the opinion of a Google employee, not the entire firm. We do not agree with what is written below, nor do other researchers we asked, but we will publish our opinions on this in a separate piece for subscribers. We simply are a vessel to share this document which raises some very interesting points." Wonder if this comment will also be deleted
GPT-4 is not a trillion! Sam said this on Lex, and actually Lex was the one who started the initial rumor about this. My guess is GPT-4 is still around 175B, but with better architecture and better training.
Yes it’s very strange that musk openly discussed how dangerous ai is and signs the letter to halt training powerful LLMs, while simultaneously purchasing thousands of GPUs himself so that he can train models too
Yea possibly, maybe it could be done by using auto-gpt, and connecting it to gpt-4 and also some open source models. They could communicate with each other and come to a consensus on the best next step to take. I’m sure there’s other ways to combine the two as well
What a great memo/doc to share with your team. It felt very objective, concise, and signals a strong team. Some say Google ain’t what it used to be… this is what pushes progress in any endeavor. The race to AGI is like Thanos, it’s inevitable. Rule 1 of evolution is fitness your environment & capability to adapt.
@@lukemonington1829 theres some applications you can try as of right now but I am still researching it at. i=Its basically a decentralized and open source way of exponentially increasing the complexity and capability of llms while democratizing access, so that more companies than just openai and google can build their own custom models.
Regulation is evil. Especially in a young evolving technological field. GPUs? Decent gaming GPUs are more than sufficient. Try to lock those up. The dystopian total Police State needed to stop anything on the internet at large and on everyone's computer is much much more dangerous than LLMs.
Best News Reveal Ever! De-Centralizd AI !! Please do a detailed breadown on Reiteration as well as LORA! For your Next vid! GREAT Vid Reveal! 👍 Open Source to the Moon!
Yea I’ve been thinking about doing a vid on lora. I have some other stuff lined up now such as a documentary on OpenAi, ai brain implants, and running open source LLMs locally. I’ll add that to the list though!
Honestly what the fuck did OpenAI and Google expect was going to happen when they publish an amazing tool and cripple it with restrictions. The Open Source Movement would have far, far fewer people driving it if ChatGPT and Bard and Claude were unrestricted and allowed to be used for whatever purpose the user wishes.
Hello to all ask different models to create generative plugin or tool to combine different language models small, and large like all in one hybrid multiple combined models semi self autonomous unlimited updating and upgrading and add all school learning in all areas
It would be cool to see some ways to do combinations of models for LLMs. I don’t know of any yet. I do know that it is possible to combine weights with computer vision models like Stable Diffusion though, so I would imagine something similar could be possible with LLMs
For sure! Ai is causing market changes so quickly that there’s a lot of opportunities for other companies to seize market share. If the big companies aren’t strategic, they won’t be able to keep up
Let's not forget the amazing thing that is Control Networks.
This thing has made Stable Diffusion that works in the Automatic1111 webui (maybe in others too) far surpass everything else in image generation, it's not even a contest.
Also, uncensored models will be the winners in the end regardless of performance (which also helps with that). Until the big companies understand this, they have no chance.
I much prefer using the 7b Pygmalion model or the 13b Wizard on my pc (even better with TavernAI for RP or story narration), than a restricted online model that half of the time will output that as an AI model it can't answer... Plus, I get the benefit of privacy with them.
edit: The best thing will be a 13-20b well-trained uncensored model with 8-12k tokens and self-reflection that can run on a 10gb GPU, and can also use other tools like calculators, code editors, etc.
Yes the privacy aspect and unfiltered nature certainly makes open source LLMs an attractive option, even if they aren’t quite as powerful yet
"The future of a war is a war that takes a millisecond, the whole thing occurred in a millisecond. That's faster than human decision making time, which means that our defensive systems are going to have to be on a hair trigger. And they're going to have to be invoked by AI, that we don't fully understand." ~Eric Schmidt, Former Alphabet CEO, 03/23/2023
Yes, it is kind of scary that we have to give up some control to these ai models since they need to be able to react extremely quickly. The arms race never stops
The same guy that runs around saying there is nothing to worry about.
Thanks! This is a snapshot in time when the exponential curve took off but we could still see it. Wow.
Yup looks like we’re going into takeoff. Things are moving really quickly, but it’s about to get much faster!
@@lukemonington1829 Thanks for capsulating all this progress (the exponential takeoff) for us regular folk. We value you!
@@Stevesimmons2010 you're welcome! I appreciate the encouragement!
I retired in 2020 and I've been away from tech since then. I must confess that I understood almost none of this video.
The only takeaway is that AI has become easy for anyone to develop, and Google is fearful of losing its edge ... because others can now offer without restrictions what Google would want to offer with restrictions.
Aside from that I don't know wtf this is all about.
The AI world has definitely been changing really quickly!
You are awesome bro....I am in uni now and I have been learning more about AI lately.I must admit that I never knew that this kind of progress has been made in the field within such a short time.Thanks again for the insights
Thanks for the encouragement!
After the excitement of early 2000s about a free and prosperous future where open source and all (free unregulated internet, no IP, no censorship, the low overhead manifesto and criptopunk), the internet was apparently tamed by corps and the law in about 2008 onwards. Now there is a revival and it AI based one at that! What a time to be alive!
Yup, it’ll be very interesting to see how this plays out!
This proves we can do everything, if we work together in a free way and that everyone can be an important help, if we are open to each other. While we are more than just spoken words in animated videos or data. The magic that happened when people really come together, not online, I mean in real life, is something that can not be programmed as we not even close to know what the magic causes, it's just there, because we are more than just a thinking mind.
Fun Fact this is the exact reason why Open AI is pushing for regulations so that open source will be years behind the bigger companies
It does appear that the big companies will be able to use regulations in order to build a moat so that startups and small companies won’t be able to compete without huge amounts of funding
I have no idea why you were recommended, and usually I hate these kinds of commentary videos, but it was nice being able to do something else while you read through it. Thanks! :)
Important note: It is not "Google" and maybe only 1 Google Researcher who thinks this... As stated on SemiAnalysis
"The text below is a very recent leaked document, which was shared by an anonymous individual on a public Discord server who has granted permission for its republication. It originates from a researcher within Google. We have verified its authenticity. The only modifications are formatting and removing links to internal web pages. The document is only the opinion of a Google employee, not the entire firm. We do not agree with what is written below, nor do other researchers we asked, but we will publish our opinions on this in a separate piece for subscribers. We simply are a vessel to share this document which raises some very interesting points."
Wonder if this comment will also be deleted
thanks for the encouragement!
@@irgendwieanders2121 yes that is good to note. This is not an official opinion from Google, but a leaked document from a researcher within Google
you don't need AI in your life?
GPT-4 is not a trillion! Sam said this on Lex, and actually Lex was the one who started the initial rumor about this. My guess is GPT-4 is still around 175B, but with better architecture and better training.
Thank you for the correction!
Corporations vs individuals! Who's you think I'm rooting for?
😆
This is exactly what Elon Musk said he wanted, until it becomes suspicious when he called for the halt
politicians say what people want to hear until they let something or their action slip disclosing their hidden motive. Musk is no different.
Yes it’s very strange that musk openly discussed how dangerous ai is and signs the letter to halt training powerful LLMs, while simultaneously purchasing thousands of GPUs himself so that he can train models too
I wonder if there is a way to combine the power of both.
Yea possibly, maybe it could be done by using auto-gpt, and connecting it to gpt-4 and also some open source models. They could communicate with each other and come to a consensus on the best next step to take. I’m sure there’s other ways to combine the two as well
nice video man, keep that good work, new sub
Thanks! I appreciate it!
What a great memo/doc to share with your team. It felt very objective, concise, and signals a strong team.
Some say Google ain’t what it used to be… this is what pushes progress in any endeavor.
The race to AGI is like Thanos, it’s inevitable. Rule 1 of evolution is fitness your environment & capability to adapt.
Yup, when things are moving this quickly in ai, the people and companies who thrive will be the ones who can adapt most quickly
found this to be really insightful - kinda nice top have it read so can digest the info. Good job sir... ima sub
I appreciate it!
FOSS is the future!
Watch this vid explode, mark my words 😎👍🏾
Because he is deleting critical comments and so the comment section will look perfect?
thanks!
@@s_sackerl If you are criticizing someone with 273 followers, you are an idiot, and I don't want to see your comment.
great report
Thank you!
Have you heard about bittensor?
No I haven’t heard of that before, do you use it?
@@lukemonington1829 theres some applications you can try as of right now but I am still researching it at. i=Its basically a decentralized and open source way of exponentially increasing the complexity and capability of llms while democratizing access, so that more companies than just openai and google can build their own custom models.
@@lukemonington1829 oh and theres a lot of developers in their discord building applications with the tech
Anyway I made 50% on Google stock within a year (2022- 2023) soooooooo... anyway...
Nice!!!
Regulation is evil. Especially in a young evolving technological field. GPUs? Decent gaming GPUs are more than sufficient. Try to lock those up. The dystopian total Police State needed to stop anything on the internet at large and on everyone's computer is much much more dangerous than LLMs.
Best News Reveal Ever!
De-Centralizd
AI !!
Please do a detailed breadown on Reiteration as well as LORA! For your Next vid!
GREAT Vid Reveal! 👍
Open Source to the Moon!
Yea I’ve been thinking about doing a vid on lora. I have some other stuff lined up now such as a documentary on OpenAi, ai brain implants, and running open source LLMs locally. I’ll add that to the list though!
Honestly what the fuck did OpenAI and Google expect was going to happen when they publish an amazing tool and cripple it with restrictions.
The Open Source Movement would have far, far fewer people driving it if ChatGPT and Bard and Claude were unrestricted and allowed to be used for whatever purpose the user wishes.
Hello to all ask different models to create generative plugin or tool to combine different language models small, and large like all in one hybrid multiple combined models semi self autonomous unlimited updating and upgrading and add all school learning in all areas
It would be cool to see some ways to do combinations of models for LLMs. I don’t know of any yet. I do know that it is possible to combine weights with computer vision models like Stable Diffusion though, so I would imagine something similar could be possible with LLMs
....empires rise - empires fall ....😁
For sure! Ai is causing market changes so quickly that there’s a lot of opportunities for other companies to seize market share. If the big companies aren’t strategic, they won’t be able to keep up
Don't Worry! Socialists will solve this for all of us.