Download the Project files for this project at my Patreon along with 250+ other projects:www.patreon.com/posts/gpt-4o-exa-multi-106754304 Learn to code fast with my 1000x MasterClass: www.patreon.com/posts/1000x-dev-103326330\ Search 200+ echohive videos and code download lnks: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
Thank you for another great course, I found it useful that we could use the chunked data to run local queries on. I also am really pleased at how regularly you post these tutorials and how relevant they are.
This is my new favorite application. I love the way that its so modular! 💜 Am getting back to the battle station tomorrow and am gonna roll this out immediately I'm applying for a new job. I think this will make a perfect scaffold for a solution that could separate me from the other candidates with some modifications. How would you suggest going about advancing this to have ensemble between multiple types of search agents (Exa vs Tavily vs Perplexity) and a "local" search from chosen database (where I can dump relevant PDFs - mainly textbooks, YT video transcripts, and research papers) These systems go from good to GREAT when adding the ensemble+synthesis layer. Ideally with easy config for how many ensemble and toggle whether to use ensemble or to single response That way its easier to control costs and tests
Thank you 🙏 glad you like it. You can check out the function which call exa api and create similar functions for other search APIs. Make sure that if the search api doesn’t return the contents of the urls then you get the content using jina ai or firecrawl or scrape them yourself. But I highly recommend you use a third party scraping service. Jina should work easily. Then you can swap out the exa ai call or use multiple search APIs simultaneously. Then chunk all results and add them to chunks.json. You would have the chunks of all search result contents in chunks.json at this and from that point on it is straight forward. How this helps.
As far as using local content check this video out: Modular Rag and chat implementation from URLs, PDFs and txt files. ruclips.net/video/E9LPaAEXjzg/видео.html This should give you ideas as I am using the same methodology almost in this project.
Download the Project files for this project at my Patreon along with 250+ other projects:www.patreon.com/posts/gpt-4o-exa-multi-106754304
Learn to code fast with my 1000x MasterClass:
www.patreon.com/posts/1000x-dev-103326330\
Search 200+ echohive videos and code download lnks: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
Thank you for another great course, I found it useful that we could use the chunked data to run local queries on. I also am really pleased at how regularly you post these tutorials and how relevant they are.
Thank you 🙏 I am happy to hear you find this and other of my projects helpful
This is my new favorite application. I love the way that its so modular! 💜
Am getting back to the battle station tomorrow and am gonna roll this out immediately
I'm applying for a new job.
I think this will make a perfect scaffold for a solution that could separate me from the other candidates with some modifications.
How would you suggest going about advancing this to have ensemble between multiple types of search agents (Exa vs Tavily vs Perplexity) and a "local" search from chosen database (where I can dump relevant PDFs - mainly textbooks, YT video transcripts, and research papers)
These systems go from good to GREAT when adding the ensemble+synthesis layer.
Ideally with easy config for how many ensemble and toggle whether to use ensemble or to single response
That way its easier to control costs and tests
Thank you 🙏 glad you like it. You can check out the function which call exa api and create similar functions for other search APIs. Make sure that if the search api doesn’t return the contents of the urls then you get the content using jina ai or firecrawl or scrape them yourself. But I highly recommend you use a third party scraping service. Jina should work easily. Then you can swap out the exa ai call or use multiple search APIs simultaneously. Then chunk all results and add them to chunks.json. You would have the chunks of all search result contents in chunks.json at this and from that point on it is straight forward. How this helps.
As far as using local content check this video out:
Modular Rag and chat implementation from URLs, PDFs and txt files.
ruclips.net/video/E9LPaAEXjzg/видео.html
This should give you ideas as I am using the same methodology almost in this project.