GPT-4o creates Knowledge trees based on any topic
Vložit
- čas přidán 18. 06. 2024
- GPT-4o creates Knowledge trees based on any topic.
Project files for this project at my Patreon along with 250+ other projects / gpt-4o-creates-104690966
Learn to code fast 1000x MasterClass: / 1000x-dev-103326330
WARNING: IF YOU MODIFY THE CODE DO NOT SET THE N_DEPTH TO MORE THAN 3, IT WILL BE MASSIVELY EXPENSIVE AND VERY QUICKLY TOO
Search 200+ echohive videos and code download links:www.echohive.live/
Auto Streamer: www.autostreamer.live/
Fastapi course: / learn-fastapi-26-95041684
Chat with us on Discord: / discord
Follow on twitter(X) : / hive_echo
CHAPTERS:
00:00 Intro
03:45 WARNING about setting n_depth!!
04:41 My Patreon benefits
05:15 Code review
#knowledgetree #gpt4o #pythonai #openaiapi #artificialintelligence #pythonprogramming #pythonprojects #largelanguagemodels
Project files at my Patreon along with 250+ other projects www.patreon.com/posts/gpt-4o-creates-104690966
Learn to code fast 1000x MasterClass: www.patreon.com/posts/1000x-dev-103326330
WARNING: IF YOU MODIFY THE CODE DO NOT SET THE N_DEPTH TO MORE THAN 3, IT WILL BE MASSIVELY EXPENSIVE AND VERY QUICKLY TOO
Search 200+ echohive videos and code download links:www.echohive.live/
Auto Streamer: www.autostreamer.live/
Fastapi course: www.patreon.com/posts/learn-fastapi-26-95041684
Chat with us on Discord: discord.gg/PPxTP3Cs3G
Follow on twitter(X) : twitter.com/hive_echo
This is awesome! I really like the frontend component. It's a nice way to explore the depth of different topics. Will be fun to play with this 😁
Thank you 🙏 other than the $160 mistake I made this was a fun project :)
🤔 Putting the output in a Knowledge Graph would be awesome❗
Yeah, that would be a nice idea
Could this be integrated into v3 autostreamer? 🤓Perhaps a way to augment the outline by editing the JSON file and then have the autostreamer iterate through everything with the max_words and such that we define.
I could see so many fun things I would want to build on top of that if it was a thing (such as adding the perplexity ensemble researcher to query each topic before populating the answer or an ensemble of OpenAI Assistants for RAG). I'd add a toggle that lets me decide which tool to use (if any)
Theoretically yes but Auto Streamer creates a more structured curriculum and this one more like a stream of consciousness style. Although, maybe I will modify this to make it more useful and less expensive so the deeper trees.
@@echohive @anydataguy i'm actually working on something highly similar that I believe i could share sources on. It uses pipes and a summarizer LLM for each topic branch and generates animations and stability video as well as some very cool text styled based on context. Is expensive though (~$20), but so cool as a demo of capability and i would say much more entertaining than perplexity. i have it filling up neo4j + Notion so you get searchable and relations too.
You always make nice videos
Thank you so much! 🙏❤️
Amazing do you know what functionality woild be amazing, if you coudld select a toic on amy level to drill down further 2 to 3 stops just on 1 name you input, does rhe same just picks one of the existing items. Also picking some for explanation would make it really cool
Thank you 🙏 and yeah I was thinking the same thing. I will try to make this more interactive. Thank you for the suggestion!
I love your videos ❤
Thank you 🙏❤️
Can you process pdfs / text files as a knowledge tree
You could in theory but it would be very expensive if done similarly in this script. This script relies on what the LLM already knows and doesn’t use any particular context. I might make a video on that separately.
@@echohive OH, ok❗ I assumed you were crawling web pages, because you mentioned "4 levels deep". A term commonly used in that Domain. Show us the Local LLM method please.
Thanks, you do good work.❗👍
Thank you 🙏 I actually tried last night with llama 3 8 billion and in lmstudio and it is very slow. I will make a video on it anyways. We can use open router to be able to make the parallel calls to OS models. I will outline both methods.
Sick
Thank you 🙏
Why do you insist on using Closed Source, instead of Open Source, over and over❓
At Least do it both ways for us who will never pay for AI. (so far )
I try to cover open source as well. I just mainly try to use the best possible models in most of my projects.
@@echohive i like i. real world wants best and open has to be aware of the bar with paid
Yeah the best models usually give more interesting results too.