Видео

This architecture is BRILLIANT and so close to my dream system/process! For the last year I've wondered: "Why hasn't anybody developed a system/app that takes API's from the top LLMs, and created agents for each... and then have these agents all work together to brainstorm, debate, review, and solve problems, and then present me with the best solution / answer that they (mostly) agree on?!?!?" I often get 4 different answers from 4 LLMs, so why not have them all setup as agents "in one room" working together to come up with the "best" solution. I can't find anybody that's tried this... why not? Wouldn't having the "top minds" (LLMs) working together produce better results? I have to imagine some very creative coders could create the logic, limits, instructions, roles, etc., to guide how the LLMs interact, balancing enough "discussion" to get to a good answer. Your architecture is so close to this!

here we go, to the moon !!! great work thanks for sharing,

Where can I get the code

I love the idea that you could aggregate predictions and then back test on past data to course correct. Thinking what else this could apply to besides trading

i tried building a simple dice game but updates kept breaking the entire gameplay. This happened a few times now. The 5 image limit is frustrating but there are ways around it.

I can save you the trouble of the making this script, by the inherent variability of the llms and the volatility of crypto/stock assets the probability drops from 45-55%(trading success) from using one to around 3-4% using 5-6 llms, A(eff)=μLLM−V. You are way better of focusing on making some kind of indicator/signal instead of a pure trader. For asset predictions lstm's or ppo's will do a much better job.

Hey Kris! I was looking for the source for this (MoPA) and am having issues locating it. Am I just overlooking it or has it not been uploaded yet?

I love this mixin❤

I am curious what would happen if you don't tell the model to be slightly negative, would the price prediction be better?

In real world bitcoin prices depend on thousands of different factors like energy price or asic hardware price, it would be much more clever to analyze those factors than simply the past prices.

I love this content. Ai news are boring. Creating tools and games with AI will be the next wave. 🤩🤩🤩

could you please confirm that to access these nim model container , do we required subscription?????

Wow, This feels like Goldmine

Hi. I’m a Dreamer. I need some money to travel the road of life. Please donate to me crypto BNB bep20: 0xAD0992311909019aC7C3A094AbAB3da99BEd70C5 Thank you.

I sincerity think this could be an amazing tool for prohrammers, specially to turn messy code into readable one 😂 then you can just check if its right

Talking to a database since the arrival of AI and LLMs doesn't seem new to me, it's a "game changer" type of use (this should be the immediate priority of all companies that to use it). I don't understand the nuance you seem to bring into this video! What do you mean with SQL, someone to explain to me in two words?

The project libukai/Awesome-ChatTTS has a more extensive documentation. It is in Chinese but Google Chrome can automatically translate at least the text elements. It doesn't replace a good english documentation but it is a good starting point.

cool stuff 😮 maybe build vs extension to smooth the workflow. seems tedious to go back and forth between claude artifact tab and vscode.

Sorry, but I find your accent very hard to understand. I'll check out Scrimba and some others. I got that part. Thank you!

Loves❤ shoutout to the air programming wizard, the master of Claude 3.5, turning code into pure poetry. Keep weaving that digital magic, bro. Your skills are next-level, turning lines of code into pure, unfiltered badassery. Stay wild, stay awesome.

Hi, friend. What kind of neuron is doing with context? I use Chatgpt 4o Plus to write code for wordpress and often I don't have enough context in order to work better. Is there more context here?

Like the visualizer idea

How can I set up a voice instruction to Clause 3.5 and it reads back to me the text it generates? Anyone has an answer for it? Something that really works. Tested and tried! THANK YOU!

How can I set up a voice instruction to Clause 3.5 and it reads back to me the text it generates? Anyone has an answer for it? Something that really works. Tested and tried! THANK YOU!

Hey Kris, been following you for a while now, thanks for everything you share. Was great to see a recording of your live stream, it just showed what a really nice person you are. Great to see good on you 👌

In the slot game it did not give full code that’s why there was still error check well

This might sound like a really dumb question, but how would one approach this when using Jan and the interface? Is that at all possible?

how to access your sourcecode as a paid channel member?

you approach using claude as your brain replacement. Maybe it's the one shot thing you are trying out. But generally you would put a little more brain effort into debugging and taking smaller steps

After watching this video and another trying to make snake: Imagine a wrapper to these feature requests that takes the generated text, compiles, sends any compile errors for rework, repeat until clean Then the human gets the results Take it a step further and have side processes unit testing the latest code, with other llms watching, making sure it behaves according to design spec

No hype just building. The best channel congrats 🎉🎉🎉❤

look good bro, thanks for always droppin bangers. Your page is worth more to me than the fuking data science degree i got.

YO! NICE

awesome job both of you 😄

where's the code and how to run this ?

Love that...a mage from Warcraft teaching AI <3

can you share the github to the above video please ? couldnt find it

Noice

Anybody remember that ball popper game? I forgot what it was even called. There was a boy that had to dodge balloons or balls, once you hit it with a blowdart it would split into 2 smaller ones and they would keep bouncing around but also lose momentum. You have to dodge the balls and pop them, but because they split into 2, you had to take care not to pop too many. I kind of want to make that game now in Android and iOS.

Does Claud have a playground where can use an api and just keep going?