Видео

Dude this is pretty cool im about to show this to a docotor i work for

Keep building🎉

Let’s go Kye! Gl on $swarms what a banger! Send some in here for more gl! Just send it, you have done it and you deserve all the attention coming your way! BuspofcKBHo2N7CjFJAy4k9Xn32WPSGw2c5UACoBT7Bm

❤

You made me lose all my money on Photon. You froze the stakes and I could not sell. Shame on you.

That's really cool kye 💪

I believe in you ky keep up the hard work and you shall prosper i will always repost all your socials 🎉

The product is very good. Can you please give me tokens? Thank you.5dgF9Wdfg8wtg3dndrMx3FsN4dCn7He9gV3FkW7i32mN

The product is very good. Can you please give me tokens? Thank you.5dgF9Wdfg8wtg3dndrMx3FsN4dCn7He9gV3FkW7i32mN

How can I use Jarvis or make one of my own ?

Do you make coin?

I'm new to investing, and l've lost a good sum trying out strategies I found in online tutorials. I would sincerely appreciate any recommendations you have.

each agent has its own llm?

🎉🎉

Very good foundational work in terms of the overall development process. However its still a very simple autonomous model. I would include more specific datasets, and apply pre existing financial models for LLMs to follow. The models work load is mainly being process by GPT, if you're going to switch to a local provider like ollama, the results are definitely going to suffer when compared to GPT. You should look into specific NLP models that are tailored to financial analysis, or train your own model on a framework like CUDA or Pytorch, which would allow you to fine tune specifically to a task. However that would require significantly more time and money. Also another good component to add, is to just allow the user to input a specific ticker in console instead of setting it inside the project.

I have been working on the concepts for doing this for crypto futures trading. But wow you guys are so far ahead.... i'm learning form 0

Did you just say git? I'm pretty sure it's pronounced git 😮😅

Very cool! 👍

Good stuff! 👍

Great content bro 👍 Liked and subscribed 💚

Super cool dude, would love to see a demo w tools being added to the agents

Hi Kye, I hope you're doing well! I wanted to check if you’re looking for a video editor and thumbnail designer for your RUclips channel. I'd love to help enhance your content and visuals. Let me know if you're interested! Best regards,

This is very interesting. I want to play with this. Thanks for putting this out there!

Let's goo

Can you do the same with mixtral or Grok, very nice contend

How to use it with Transformers? model = Qwen2VLForConditionalGeneration.from_pretrained( quant_awq, torch_dtype="auto", device_map="auto", cache_dir=model_path, # attn_implementation="flash_attention_2", ) the pip install flash-attn is not installing correctly in windows 10. CUDA 11.8 pytorch 2.2. So how to use this Zeta as alternative?

This is cool! you got a new sub

Want to try it as webui pipeline

Would you be open to a PR for a local UI?

Well done Kye🙌

Do you think you are at the phase where a ui would be good even locally or running in terminal is better?

On the right path

IS IT BEETTER THAN agency swarm agents?

How does this differ from having an "allrounder agent" with functions/tools defined for specialized tasks which are then executed by specialized agents with the right llm selection and the proper system prompt for that task ?

I dont really get why this should be true, the agents need the information or response of the one before, so you are simply chaining agent responses, i could loop that with one agent and create a task plan before. Also i checked the papers and i think they are outdated. Can you enlighten me on that ?

A video on how this framework differs or aims to differ from the other workflow / agent frameworks out there would be useful as well.

way too underrated , but ill gate keep this 😅

Cant wait to try it out :)

Looking forward to seeing tool use and RAG!

bro it is not working I was literally running the cats code from the github documentation, it gives this error FieldValidatorDecoratorInfo.__init__() got an unexpected keyword argument 'json_schema_input_type' this is the code I tried from fluid_api_agent.main import ( fluid_api_request, ) # Example 1: Basic API Request basic_request = fluid_api_request( "Generate an API request to get a random cat fact from catfact.ninja/fact" ) print(basic_request.model_dump_json(indent=4))

how did this almost take 20 min and the program wasnt even that advanced

Super neat. Nice video.

do sol

Excellent video!

coding in 720p?

The audio is a total disaster, or extremely funny.

not mentioning the fail to give credit for academic papers, it was so cringe to see the announcement of o1 the dude adult was clearly ashamed of those kids, of course they came up by themselves

cooool

How much compute does this require? I don't want to make a big llm

Really amazing 🎉